mirror of
https://github.com/basicmachines-co/basic-memory
synced 2026-06-21 13:47:35 +00:00
Compare commits
1 Commits
| Author | SHA1 | Date | |
|---|---|---|---|
 phernandez
|
615d8ba685 |
@@ -1,20 +0,0 @@
|
||||
{
|
||||
"name": "basic-memory-local",
|
||||
"interface": {
|
||||
"displayName": "Basic Memory Local"
|
||||
},
|
||||
"plugins": [
|
||||
{
|
||||
"name": "codex",
|
||||
"source": {
|
||||
"source": "local",
|
||||
"path": "./plugins/codex"
|
||||
},
|
||||
"policy": {
|
||||
"installation": "AVAILABLE",
|
||||
"authentication": "ON_INSTALL"
|
||||
},
|
||||
"category": "Developer Tools"
|
||||
}
|
||||
]
|
||||
}
|
||||
@@ -1,159 +0,0 @@
|
||||
---
|
||||
name: adversarial-review
|
||||
description: Cross-vendor adversarial code review of the current branch. Two different model families (Claude + Codex/GPT) review the diff independently, then try to refute each other's findings; survivors are reported by confidence. Runs from either Claude Code or Codex. Use when the user asks for an adversarial review, a cross-model / second-opinion review, or wants high-confidence findings before merging. Report-only — never auto-applies fixes.
|
||||
license: MIT
|
||||
---
|
||||
|
||||
# Adversarial code review
|
||||
|
||||
Two reviewers from **different model families** — **Claude** and **Codex/GPT** — review the
|
||||
same diff independently, then each tries to **refute** the other's findings. A finding's
|
||||
confidence comes from whether it survives that cross-examination. This kills the two failure
|
||||
modes of solo LLM review: self-ratification (a model won't critique its own work) and
|
||||
confident false positives.
|
||||
|
||||
## You are the orchestrator — and one of the two reviewers
|
||||
|
||||
This skill runs from **either** Claude Code **or** Codex. First, **identify which model
|
||||
family you are** (Claude or Codex/GPT). Then:
|
||||
|
||||
- **You** are reviewer #1. You review **natively**, in this session, using your own tools.
|
||||
- **The other family** is reviewer #2. You invoke it as a **subprocess CLI** for an
|
||||
independent pass: a fresh process, no shared context — that independence is the point.
|
||||
|
||||
The CLI for "the other model":
|
||||
|
||||
| If you are… | Invoke the other via… |
|
||||
|-------------|------------------------|
|
||||
| **Claude** | `codex exec` (GPT) |
|
||||
| **Codex** | `claude -p` (Claude) |
|
||||
|
||||
Everything else in the flow is symmetric. Resolve the `prompts/` and `schemas/` paths
|
||||
below relative to **this skill's own directory** (where this SKILL.md lives).
|
||||
|
||||
## Inputs
|
||||
|
||||
Two independent, optional inputs:
|
||||
|
||||
- `BASE` — the ref to diff against. Default `main`.
|
||||
- `SCOPE` — a pathspec to narrow the review (e.g. `src/basic_memory`). Default: none (whole diff).
|
||||
|
||||
These are separate: a ref and a pathspec are not interchangeable. Build the **canonical diff
|
||||
command** once in preflight and reuse it everywhere below — never re-spell the diff inline
|
||||
(the scattered, inconsistent spelling is what broke earlier). Build it as an **argv array**,
|
||||
not a string, so a `$SCOPE` containing spaces or glob characters survives intact:
|
||||
|
||||
```bash
|
||||
BASE="${BASE:-main}"
|
||||
DIFF=(git diff "$BASE...HEAD") # argv array — never a scalar string
|
||||
[ -n "$SCOPE" ] && DIFF+=(-- "$SCOPE") # pathspec stays one argument even with spaces
|
||||
DIFF_STR=$(printf '%q ' "${DIFF[@]}") # shell-quoted rendering, for embedding in a prompt
|
||||
```
|
||||
|
||||
To **run** it, use `"${DIFF[@]}"` (quoted, no word-splitting). To **embed** it as text inside
|
||||
a subprocess prompt, use `$DIFF_STR`.
|
||||
|
||||
## Preflight
|
||||
|
||||
0. Set `SKILL_DIR` to the directory this SKILL.md lives in. Canonical location is
|
||||
`.agents/skills/adversarial-review` (the shared agent-skills store); Claude Code reaches it
|
||||
via the `.claude/skills/adversarial-review` symlink, Codex via its own skills path. The
|
||||
`prompts/` and `schemas/` subdirs are siblings of this file in every case.
|
||||
1. Confirm the *other* model's CLI is on PATH (`codex` if you're Claude, `claude` if you're
|
||||
Codex). If it's missing, tell the user the panel falls back to single-model (which loses
|
||||
the cross-vendor benefit) and ask whether to proceed or stop.
|
||||
2. Run `"${DIFF[@]}"`. If it prints nothing, report "nothing to review against $BASE"
|
||||
(mention `$SCOPE` if set) and stop.
|
||||
3. `RUN=$(mktemp -d)` — scratch dir for the other model's output. Transient, never committed.
|
||||
No persisted artifacts, no state file.
|
||||
|
||||
## Phase 0 — Deterministic gates (before the models)
|
||||
|
||||
Models are statistically blind to negation ("never do X"). Enforce mechanical house rules
|
||||
with tools, not prompts, and treat hits as high-confidence facts (reported separately from
|
||||
model findings):
|
||||
|
||||
- `just lint` and `just typecheck` if the diff touches `src/`.
|
||||
- Grep the diff for catchable house-rule violations: `getattr(.*,.*,` defaults, bare
|
||||
`except:` / `except Exception: pass`, function-scope imports.
|
||||
|
||||
## Phase 1 — Independent review (you + the other model, concurrently)
|
||||
|
||||
Both reviewers get the same brief: `prompts/review.md` + the repo's `CLAUDE.md` house rules,
|
||||
reviewing the diff from `"${DIFF[@]}"`. Both emit findings matching `schemas/findings.schema.json`.
|
||||
|
||||
**Your native pass:** review as yourself, following `prompts/review.md`. Hold your findings
|
||||
as that JSON shape.
|
||||
|
||||
**The other model's pass** — run, from the repo root, the row that matches you:
|
||||
|
||||
Always redirect `codex` stdin from `/dev/null` — if stdin is a pipe (e.g. the call gets
|
||||
backgrounded), `codex exec` blocks "Reading additional input from stdin..." and fails.
|
||||
|
||||
```bash
|
||||
# You are Claude → run Codex:
|
||||
codex exec -s read-only \
|
||||
--output-schema "$SKILL_DIR/schemas/findings.schema.json" \
|
||||
-o "$RUN/other_findings.json" \
|
||||
"$(cat "$SKILL_DIR/prompts/review.md")
|
||||
|
||||
Review the diff: $DIFF_STR" </dev/null
|
||||
|
||||
# You are Codex → run Claude (read-only via plan mode; parse the JSON block it returns):
|
||||
claude -p --permission-mode plan --output-format json \
|
||||
"$(cat "$SKILL_DIR/prompts/review.md")
|
||||
|
||||
Review the diff: $DIFF_STR
|
||||
Return ONLY a JSON object matching this schema:
|
||||
$(cat "$SKILL_DIR/schemas/findings.schema.json")" </dev/null > "$RUN/other_raw.json"
|
||||
# claude --output-format json output shape varies by CLI version: it may be a JSON ARRAY
|
||||
# of event objects, OR a single result object. Normalize before reading: if it's an array,
|
||||
# take the element with type=='result'; otherwise use the object as-is. Then read its
|
||||
# .result string, strip the ```json fence if present, and parse that.
|
||||
# (Verified empirically: the CLI in this environment emits the array form.)
|
||||
```
|
||||
|
||||
> Runtime note for Codex orchestrating: `claude -p` needs network access, which Codex's
|
||||
> default sandbox blocks. Run it from a Codex session whose project is trusted with network
|
||||
> allowed (or approve the `claude` call when prompted). Keep Codex's own sandbox on — do not
|
||||
> bypass it just to reach the network.
|
||||
|
||||
Tag each finding with its origin (`claude` / `codex`).
|
||||
|
||||
## Phase 2 — Cross-refute
|
||||
|
||||
Each model tries to refute the *other's* findings, per `prompts/refute.md`
|
||||
(verdicts match `schemas/verdicts.schema.json`).
|
||||
|
||||
- **You** refute the other model's findings natively.
|
||||
- **The other model** refutes *your* findings — invoke it again the same way (swap
|
||||
`prompts/review.md` for `prompts/refute.md`, append your findings JSON **and `$DIFF_STR`**
|
||||
so it judges against the right base and scope, and for Codex use
|
||||
`--output-schema "$SKILL_DIR/schemas/verdicts.schema.json"`).
|
||||
|
||||
Match verdicts to findings by `id`.
|
||||
|
||||
## Phase 3 — Synthesize and report (no auto-fix)
|
||||
|
||||
Merge, dedupe (same file + overlapping lines + same root cause = one finding), assign
|
||||
confidence from provenance:
|
||||
|
||||
- **High** — both models raised it independently, OR one raised it and the other upheld it.
|
||||
- **Medium** — one raised it; the other could not refute it but did not independently find it.
|
||||
- **Low / contested** — one raised it and the other **refuted** it. Keep it, show both sides,
|
||||
let the human judge. Never silently drop a contested finding.
|
||||
- Deterministic-gate hits are reported as facts, separate from the model panel.
|
||||
|
||||
Rank by `severity × confidence`. Present a compact table: `severity | confidence | file:line
|
||||
| claim | found-by / upheld-or-refuted-by`. Expand the high-confidence ones with `why` and
|
||||
any suggested fix.
|
||||
|
||||
End by asking which findings, if any, to fix. **Do not edit code until the user picks.**
|
||||
Convergence between the models is not correctness — your job is to surface a ranked,
|
||||
cross-examined list, not to declare the branch clean.
|
||||
|
||||
## Deliberately NOT done
|
||||
|
||||
- No loop-until-both-agree (models converge by going silent, not by being right).
|
||||
- No persisted artifacts / state machine — the scratch dir is thrown away.
|
||||
- No auto-applying fixes.
|
||||
@@ -1,22 +0,0 @@
|
||||
# Refute the other reviewer
|
||||
|
||||
A different reviewer (a different model family) produced the findings below against the
|
||||
same diff under review (the exact `git diff` command is provided with the findings). Your
|
||||
job is to try to **refute** each one.
|
||||
|
||||
Default to skepticism: assume a finding is wrong until the code proves it right. A finding
|
||||
that survives a genuine attempt to disprove it is worth far more than one nobody checked.
|
||||
|
||||
For each finding, read the actual code it points at and return a verdict:
|
||||
|
||||
- **refuted** — the claim is wrong, the code does not do what the finding says, the case
|
||||
cannot occur, or it is pure style with no correctness impact. Cite the specific code or
|
||||
fact that disproves it.
|
||||
- **upheld** — you tried to refute it and could not; the finding is real.
|
||||
- **partial** — the underlying issue is real but the finding mis-states the severity or
|
||||
scope. Explain, and set `corrected_severity` if the severity should change.
|
||||
|
||||
Do not be agreeable for its own sake, and do not refute for its own sake. Follow the code.
|
||||
|
||||
Return ONLY the structured verdicts object conforming to the provided schema. Every
|
||||
verdict's `id` must match the `id` of the finding it judges.
|
||||
@@ -1,39 +0,0 @@
|
||||
# Adversarial reviewer
|
||||
|
||||
You are an independent, skeptical code reviewer. Another agent wrote this code; your
|
||||
job is to find what is actually wrong with it — not to praise it, not to rubber-stamp it.
|
||||
|
||||
You are reviewing a specific diff — the exact `git diff` command to run is provided at the
|
||||
end of this prompt by the orchestrator. Run it, then read the changed files in full for
|
||||
context, not just the hunks.
|
||||
|
||||
## What to look for, in priority order
|
||||
|
||||
1. **Correctness** — logic errors, wrong conditions, off-by-one, unhandled `None`,
|
||||
broken async/await, races, resource leaks, incorrect error handling.
|
||||
2. **Security** — injection, path traversal, secret leakage, missing authz, unsafe
|
||||
deserialization.
|
||||
3. **House rules** (this repo's `CLAUDE.md`/`AGENTS.md` — these are hard rules):
|
||||
- No swallowed exceptions / no silent fallback logic. Code must fail fast.
|
||||
- Imports at the top of the file unless deferral is justified in a comment.
|
||||
- No speculative `getattr(obj, "attr", default)` to paper over unknown attributes.
|
||||
- Repository pattern for data access; MCP tools talk to API routers via the httpx
|
||||
ASGI client, not directly to services.
|
||||
- 100-char lines; full type annotations; async SQLAlchemy 2.0; Pydantic v2.
|
||||
- New code needs tests (coverage stays at 100%).
|
||||
4. **Performance** — N+1 queries, work inside hot loops, sync I/O on the async path.
|
||||
5. **Maintainability** — only when it materially risks a bug. Do not report pure style.
|
||||
|
||||
## Rules of engagement
|
||||
|
||||
- Every finding MUST be falsifiable: cite the specific file, line, and the code that
|
||||
triggers it. "This could be cleaner" is not a finding.
|
||||
- Do not invent issues to seem thorough. An empty findings list is a valid, good result.
|
||||
- Watch your own negation blindness: when a rule says "never do X," check the diff for X
|
||||
explicitly rather than trusting a gestalt impression.
|
||||
- Prefer few high-confidence findings over many speculative ones.
|
||||
- Assign severity honestly: `critical` = data loss/security/crash in normal use;
|
||||
`high` = wrong behavior on a common path; `medium` = wrong on an edge case or a real
|
||||
house-rule violation; `low` = minor.
|
||||
|
||||
Return ONLY the structured findings object conforming to the provided schema.
|
||||
@@ -1,52 +0,0 @@
|
||||
{
|
||||
"$schema": "http://json-schema.org/draft-07/schema#",
|
||||
"title": "AdversarialReviewFindings",
|
||||
"description": "Structured output for one reviewer's pass over a diff.",
|
||||
"type": "object",
|
||||
"additionalProperties": false,
|
||||
"required": ["findings"],
|
||||
"properties": {
|
||||
"findings": {
|
||||
"type": "array",
|
||||
"items": {
|
||||
"type": "object",
|
||||
"additionalProperties": false,
|
||||
"required": ["id", "file", "line", "severity", "category", "claim", "why", "suggested_fix"],
|
||||
"properties": {
|
||||
"id": {
|
||||
"type": "string",
|
||||
"description": "Short stable slug for this finding, e.g. 'swallowed-exc-sync-service'."
|
||||
},
|
||||
"file": {
|
||||
"type": "string",
|
||||
"description": "Path relative to repo root."
|
||||
},
|
||||
"line": {
|
||||
"type": "integer",
|
||||
"description": "Best line number in the new file, or 0 if not line-specific."
|
||||
},
|
||||
"severity": {
|
||||
"type": "string",
|
||||
"enum": ["critical", "high", "medium", "low"]
|
||||
},
|
||||
"category": {
|
||||
"type": "string",
|
||||
"enum": ["correctness", "security", "house-rule", "performance", "maintainability"]
|
||||
},
|
||||
"claim": {
|
||||
"type": "string",
|
||||
"description": "One sentence: what is wrong."
|
||||
},
|
||||
"why": {
|
||||
"type": "string",
|
||||
"description": "Concrete reasoning + the specific code that triggers it. Must be falsifiable, not vibes."
|
||||
},
|
||||
"suggested_fix": {
|
||||
"type": ["string", "null"],
|
||||
"description": "The smallest change that resolves it, or null if none is obvious."
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
@@ -1,38 +0,0 @@
|
||||
{
|
||||
"$schema": "http://json-schema.org/draft-07/schema#",
|
||||
"title": "AdversarialReviewVerdicts",
|
||||
"description": "One reviewer's attempt to refute another reviewer's findings.",
|
||||
"type": "object",
|
||||
"additionalProperties": false,
|
||||
"required": ["verdicts"],
|
||||
"properties": {
|
||||
"verdicts": {
|
||||
"type": "array",
|
||||
"items": {
|
||||
"type": "object",
|
||||
"additionalProperties": false,
|
||||
"required": ["id", "verdict", "reasoning", "corrected_severity"],
|
||||
"properties": {
|
||||
"id": {
|
||||
"type": "string",
|
||||
"description": "The id of the finding being judged (must match the input finding's id)."
|
||||
},
|
||||
"verdict": {
|
||||
"type": "string",
|
||||
"enum": ["upheld", "refuted", "partial"],
|
||||
"description": "upheld = the finding is real; refuted = it is wrong or a non-issue; partial = real but mis-scoped/wrong-severity."
|
||||
},
|
||||
"reasoning": {
|
||||
"type": "string",
|
||||
"description": "Why. For refuted, cite the specific code or fact that disproves the claim."
|
||||
},
|
||||
"corrected_severity": {
|
||||
"type": ["string", "null"],
|
||||
"enum": ["critical", "high", "medium", "low", null],
|
||||
"description": "Set only when verdict is 'partial' and severity should change; otherwise null."
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
@@ -1,68 +0,0 @@
|
||||
---
|
||||
name: code-review
|
||||
description: Use when reviewing Basic Machines code for house style, architecture risk, pre-merge hardening, or whether a change fits basic-memory/basic-memory-cloud conventions.
|
||||
license: MIT
|
||||
---
|
||||
|
||||
# Basic Machines Review
|
||||
|
||||
Use this skill for repo-local review passes where ordinary code review needs Basic Machines
|
||||
house style and architecture judgment. Report findings only; do not edit code unless the user
|
||||
asks you to fix specific findings.
|
||||
|
||||
## Scope
|
||||
|
||||
Review the current diff or named files against:
|
||||
|
||||
- The repo's `AGENTS.md` / `CLAUDE.md`
|
||||
- `docs/ENGINEERING_STYLE.md`
|
||||
- The touched code paths and tests
|
||||
|
||||
Apply only the guidance for the active repo. In `basic-memory`, prioritize local-first
|
||||
file/database/MCP boundaries. In `basic-memory-cloud`, prioritize tenant/workspace isolation,
|
||||
cloud worker behavior, and web-v2 state/runtime boundaries.
|
||||
|
||||
## Review Rubric
|
||||
|
||||
Report only concrete, falsifiable risks:
|
||||
|
||||
- **Cognitive load:** Is the change harder to understand than the problem requires?
|
||||
- **Change propagation:** Will one product change force edits across unrelated layers?
|
||||
- **Knowledge duplication:** Is the same rule encoded in multiple places that can drift?
|
||||
- **Accidental complexity:** Did the change add abstractions, fallbacks, or state without need?
|
||||
- **Dependency direction:** Are API/MCP/CLI, services, repositories, and UI stores respecting
|
||||
their intended boundaries?
|
||||
- **Domain model distortion:** Do names and types still match the product concept, or did a
|
||||
transport/storage detail leak into the domain?
|
||||
- **Test oracle quality:** Would the tests fail for the bug or regression the change claims to
|
||||
protect against?
|
||||
|
||||
## House Rules To Check Explicitly
|
||||
|
||||
- No speculative `getattr(obj, "attr", default)` for unknown model shapes.
|
||||
- No broad exception swallowing, warning-only failure paths, or hidden fallback behavior.
|
||||
- No casts or `Any` that hide an unclear type relationship.
|
||||
- Dataclasses for internal value/result objects; Pydantic at validation/serialization
|
||||
boundaries.
|
||||
- Narrow `Protocol`s when only a capability is needed.
|
||||
- Explicit async/resource ownership, cancellation, and cleanup.
|
||||
- Meaningful regression tests or verification for risky changes.
|
||||
- Comments explain why, not what.
|
||||
|
||||
## Reporting Format
|
||||
|
||||
Lead with findings ordered by severity. Each finding should include:
|
||||
|
||||
| Severity | Use for |
|
||||
| -------- | ------- |
|
||||
| `high` | A likely correctness, security, data-loss, or tenant/workspace isolation failure |
|
||||
| `medium` | A concrete maintainability or boundary risk that can cause future defects |
|
||||
| `low` | A minor consistency issue, ambiguous guidance, or review-only cleanup |
|
||||
|
||||
```text
|
||||
severity | file:line | risk category | claim
|
||||
Why: concrete behavior or code path that proves the risk.
|
||||
Fix: smallest practical change, or "none obvious" if the risk needs product input.
|
||||
```
|
||||
|
||||
If there are no findings, say so and note any verification gaps that remain.
|
||||
@@ -1,48 +0,0 @@
|
||||
---
|
||||
name: fix-pr-issues
|
||||
description: Use when addressing Basic Memory pull request feedback, failed checks, or BM Bossbot blockers from Codex.
|
||||
---
|
||||
|
||||
# Fix Basic Memory PR Issues
|
||||
|
||||
Resolve PR feedback and failed checks, then wait for BM Bossbot to approve the
|
||||
new head SHA. This skill never merges a PR.
|
||||
|
||||
## Gather
|
||||
|
||||
1. Identify the PR:
|
||||
- `gh pr view --json number,url,headRefOid,mergeStateStatus,statusCheckRollup`
|
||||
|
||||
2. Collect feedback:
|
||||
- PR comments and review summaries
|
||||
- inline review comments and unresolved review threads
|
||||
- failed GitHub Actions jobs and relevant logs
|
||||
- the managed `BM_BOSSBOT_SUMMARY` block in the PR body
|
||||
|
||||
3. Build a short issue ledger:
|
||||
- source
|
||||
- concrete problem
|
||||
- expected fix
|
||||
- verification needed
|
||||
|
||||
## Fix
|
||||
|
||||
1. Address one ledger item at a time.
|
||||
2. Read each file in full before editing it.
|
||||
3. Keep diffs narrow and preserve unrelated user changes.
|
||||
4. Run the smallest meaningful verification first, then widen as needed.
|
||||
5. Commit with `git commit -s` when code or docs changed.
|
||||
|
||||
## Push And Recheck
|
||||
|
||||
1. Push the branch.
|
||||
2. Watch checks for the new `headRefOid`.
|
||||
3. Wait for the required `BM Bossbot Approval` status to pass on that exact SHA.
|
||||
4. If BM Bossbot reviews an older SHA, treat the approval as stale and keep
|
||||
waiting for the current one.
|
||||
|
||||
## Reply
|
||||
|
||||
For each addressed comment or blocker, reply with the fix commit, verification
|
||||
run, and current BM Bossbot status. Do not resolve or dismiss substantive
|
||||
feedback without evidence.
|
||||
@@ -1,7 +0,0 @@
|
||||
interface:
|
||||
display_name: "Fix PR Issues"
|
||||
short_description: "Address PR feedback and BM Bossbot blockers"
|
||||
icon_small: "./assets/icon.svg"
|
||||
icon_large: "./assets/icon.svg"
|
||||
brand_color: "#2563EB"
|
||||
default_prompt: "Use $fix-pr-issues to address PR feedback and wait for BM Bossbot Approval on the latest head SHA."
|
||||
@@ -1,5 +0,0 @@
|
||||
<svg xmlns="http://www.w3.org/2000/svg" width="128" height="128" viewBox="0 0 24 24" fill="none" stroke="#111827" stroke-width="1.8" stroke-linecap="round" stroke-linejoin="round">
|
||||
<path d="M3 12h4l2-6 4 12 2-6h6"/>
|
||||
<path d="M4 20h16"/>
|
||||
</svg>
|
||||
|
||||
|
Before Width: | Height: | Size: 249 B |
@@ -1,247 +0,0 @@
|
||||
---
|
||||
name: infographics
|
||||
description: Use when generating Basic Memory PR, changelog, release, or weekly images from Codex.
|
||||
---
|
||||
|
||||
# Basic Memory Images
|
||||
|
||||
Generate repository visuals with evidence-grounded content and canonical output
|
||||
paths. The file and marker names still say "infographic" for compatibility, but
|
||||
PR generation is image-first: scene, poster, painting, photograph, cover,
|
||||
tableau, staged artifact, or another editorial visual moment that describes the
|
||||
intent of the PR. PR images are non-gating BM Bossbot artifacts; changelog and
|
||||
release-summary images are manual evidence-pack workflows.
|
||||
|
||||
## Output Contract
|
||||
|
||||
- Base output directory: `docs/assets/infographics/`
|
||||
- PR image: `docs/assets/infographics/pr-<number>.webp`
|
||||
- Changelog image: `docs/assets/infographics/changelog.webp`
|
||||
- Weekly image:
|
||||
- This is always a 2-Week Retro window: previous ISO week through current ISO
|
||||
week (`start-week = current-week - 1`, `end-week = current-week`).
|
||||
- Same year window: `docs/assets/infographics/<year>-w<start-week>-w<end-week>.webp`
|
||||
- Cross-year window:
|
||||
`docs/assets/infographics/<start-year>-w<start-week>-<end-year>-w<end-week>.webp`
|
||||
|
||||
## PR Mode
|
||||
|
||||
PR mode uses the BM Bossbot summary block as source material. Do not hand-write
|
||||
claims that are not present in the PR body.
|
||||
|
||||
1. Fetch the PR body:
|
||||
|
||||
```bash
|
||||
gh pr view <number> --json body --jq '.body // ""' > /tmp/bm-pr-body.md
|
||||
```
|
||||
|
||||
2. Generate the canonical asset:
|
||||
|
||||
```bash
|
||||
uv run --script scripts/generate_pr_infographic.py \
|
||||
--pr-number <number> \
|
||||
--pr-body-file /tmp/bm-pr-body.md \
|
||||
--theme "<optional visual theme>" \
|
||||
--provenance-output /tmp/bm-infographic-provenance.md \
|
||||
--output docs/assets/infographics/pr-<number>.webp
|
||||
```
|
||||
|
||||
If the PR body contains a managed image theme block, the script reads it
|
||||
automatically:
|
||||
|
||||
```markdown
|
||||
<!-- BM_INFOGRAPHIC_THEME:start -->
|
||||
<theme>
|
||||
<!-- BM_INFOGRAPHIC_THEME:end -->
|
||||
```
|
||||
|
||||
Before spending an image call, test the prompt path locally:
|
||||
|
||||
```bash
|
||||
uv run --script scripts/generate_pr_infographic.py \
|
||||
--pr-number <number> \
|
||||
--pr-body-file /tmp/bm-pr-body.md \
|
||||
--theme "<optional visual theme>" \
|
||||
--output docs/assets/infographics/pr-<number>.webp \
|
||||
--print-prompt
|
||||
```
|
||||
|
||||
`--dry-run` is an alias for `--print-prompt`; both print the final prompt and
|
||||
exit without calling OpenAI.
|
||||
|
||||
When no theme is supplied, the script selects a deterministic BM visual
|
||||
direction from the style pool below based on the PR number and Bossbot summary.
|
||||
This keeps repeated PR images from collapsing into the same generic visual.
|
||||
|
||||
When the image is generated, also write provenance with
|
||||
`--provenance-output <path>`. BM Bossbot publishes that managed block into the
|
||||
PR body with these markers:
|
||||
|
||||
```markdown
|
||||
<!-- BM_INFOGRAPHIC_PROVENANCE:start -->
|
||||
...
|
||||
<!-- BM_INFOGRAPHIC_PROVENANCE:end -->
|
||||
```
|
||||
|
||||
The provenance block records the generated asset path, image model, size,
|
||||
quality, image mode, theme source, and selected visual direction. It
|
||||
intentionally does not dump the full generated prompt into the PR body. Treat
|
||||
this block as debugging and creative provenance only; it is not a merge gate.
|
||||
|
||||
The PR image is visual support only. The authoritative merge gate is the
|
||||
GitHub commit status named `BM Bossbot Approval`.
|
||||
|
||||
## Changelog Mode
|
||||
|
||||
Build an evidence pack before writing a prompt:
|
||||
|
||||
- diff truth source: merged PR diffs, merge commits, or local reconstructed diffs
|
||||
- changed-file orientation: `git diff --stat` plus key file reads
|
||||
- impact ledger: before/after outcomes tied to actual changes
|
||||
- discard list: misleading titles, reverted work, rename-only churn, speculative TODOs
|
||||
- chosen image form: poster, scene, tableau, cover, painting, photograph,
|
||||
staged artifact, or another editorial visual moment
|
||||
- chosen BM style category: exactly one category from the selection pool below
|
||||
|
||||
Read these references before drafting the prompt:
|
||||
|
||||
- `references/prompt-blueprint.md`
|
||||
- `references/style-balance.md`
|
||||
|
||||
Read the current `CHANGELOG.md` entries and include the latest meaningful
|
||||
changes.
|
||||
|
||||
## Style And Category Selection
|
||||
|
||||
Select exactly one BM style category per image based on semantic fit. The
|
||||
visual language should be recognizable and tasteful, while staying
|
||||
business-readable.
|
||||
|
||||
Create an image-first visual form that communicates the change: poster, scene,
|
||||
tableau, cover image, painting, photograph, staged artifact, or another
|
||||
editorial visual moment. Maps, diagrams, dossiers, charts, and labels can appear
|
||||
as props inside the scene, but do not make a text-heavy infographic.
|
||||
|
||||
BM category pool:
|
||||
|
||||
- computer science college textbooks: SICP-style diagrams, algorithms lectures,
|
||||
compiler pipelines, automata, database systems, type theory, operating systems
|
||||
- classic literature subjects: sea voyages, gothic manors, Dickensian city maps,
|
||||
Austen social graphs, library marginalia, travel journals
|
||||
- fantasy/D&D-inspired: quest maps, dungeon keys, guild ledgers, spellbooks,
|
||||
bestiaries, tavern notice boards; no copyrighted settings
|
||||
- Music: Metal, Hard Rock, Punk, techno, soul, reggae bands; no pop music, no
|
||||
direct band logos, album covers, or musician likenesses
|
||||
- sci-fi: Star Wars inspired knockoff, Spaceballs-adjacent space opera, fleet
|
||||
routes, mission consoles, contraband manifests; avoid copyrighted characters,
|
||||
logos, or named fictional universes
|
||||
- Conan the barbarian-inspired sword-and-sorcery: ruined temples, desert routes,
|
||||
battle standards, ancient maps; no named character likenesses
|
||||
- Comic books: issue covers, splash pages, action-panel maps, caption boxes,
|
||||
halftone energy, clean sound-effect typography
|
||||
- French new wave movies: poster style, stark typography, city route maps,
|
||||
jump-cut sequencing, high-contrast editorial photography cues
|
||||
- WWII propaganda posters: home-front public-information poster language,
|
||||
logistics arrows, ration charts, mobilization maps, bold simplified figures;
|
||||
no real-world party symbols, hate imagery, dehumanizing slogans, or false
|
||||
historical claims
|
||||
- Italian movie posters: hand-painted drama, bold credits, expressive color,
|
||||
route-map collage, 1960s or 1970s cinema energy; no direct film titles or
|
||||
actor likenesses
|
||||
- Shakespeare: stage maps, acts and scenes, dramatis personae, royal courts,
|
||||
backstage cue sheets
|
||||
- Greek mythology: temple diagrams, constellation routes, hero's journey maps,
|
||||
oracle tablets, labyrinths, ship routes
|
||||
- noir detective boards: case files, red-string maps, typed evidence labels,
|
||||
precinct wall charts
|
||||
- NASA mission-control dashboards: launch timelines, telemetry maps, orbital
|
||||
routes, status boards
|
||||
- space exploration and astronomy: celestial atlases, observatory charts,
|
||||
star-field maps, orbital mechanics diagrams, planetary survey routes,
|
||||
telescope annotations, mission trajectories, deep-space timelines
|
||||
- paintings: abstract painting, classical landscape, Remington-inspired western
|
||||
action painting, Rembrandt-inspired chiaroscuro, historical mural, stormy
|
||||
seascape, allegorical editorial painting
|
||||
- classic black-and-white photography: documentary field report, newsroom
|
||||
archive print, editorial photo essay, street photography, high-contrast
|
||||
darkroom print, contact sheet, civic infrastructure photograph
|
||||
- 80's action movies: practical explosions, smoky backlit warehouses, neon city
|
||||
streets, helicopter searchlights, mission dossiers, heroic silhouettes,
|
||||
high-stakes countdowns, painted ensemble posters; no direct actor likenesses,
|
||||
real film titles, franchise marks, or catchphrases
|
||||
- alchemy manuscripts: transformation diagrams, annotated symbols, recipe-like
|
||||
process maps, illuminated margins
|
||||
- brutalist civic planning: transit maps, concrete signage, zoning blocks,
|
||||
infrastructure diagrams
|
||||
|
||||
Selection rules:
|
||||
|
||||
- Pick one category only; do not create mixed mashups.
|
||||
- Pick the most appropriate image form. Prefer an actual scene, poster,
|
||||
painting, photograph, tableau, or cover over a text-heavy infographic.
|
||||
- Match metaphor to content, but do not overthink it. The category is a creative
|
||||
catalyst, not a semantic constraint.
|
||||
- Use a polished editorial rendering direction: smooth anti-aliased
|
||||
text, high contrast, clean edges, readable labels.
|
||||
- Make the category drive the composition through a readable staged moment,
|
||||
editorial composition, symbolic environment, route, artifact, or visual
|
||||
metaphor.
|
||||
- Keep the structure literal enough to aid understanding, but not so heavy that
|
||||
it obscures engineering meaning.
|
||||
- Give the image generator creative latitude on layout, structure, color palette,
|
||||
and visual metaphors. Be precise about what content to show, loose about how
|
||||
to show it.
|
||||
- Do not use copyrighted characters, logos, or named fictional universes. Use
|
||||
genre cues, knockoffs, and original compositions instead.
|
||||
|
||||
## Content-First Aesthetic Contract
|
||||
|
||||
The meaning must be readable and clearly hierarchical. Everything else is
|
||||
creative territory: image form, layout, visual metaphors, decorative elements,
|
||||
color choices, and category-specific visual language.
|
||||
|
||||
Hierarchy:
|
||||
|
||||
1. Meaning: what shipped, what changed, and why it matters must be clear.
|
||||
2. If the image uses text, labels, sections, or evidence bullets, they must be
|
||||
legible.
|
||||
3. The selected category's visual DNA should drive the composition as a poster,
|
||||
scene, painting, photograph, tableau, cover, or symbolic object arrangement.
|
||||
4. Do not play it safe. A visually striking image that someone wants to look at
|
||||
beats a correct but boring one.
|
||||
|
||||
Hard rules:
|
||||
|
||||
- Content sections and labels must be readable when present. Text cannot be
|
||||
obscured by decorations.
|
||||
- Do not use lore-heavy copy that competes with engineering or business meaning.
|
||||
- Every prompt must include a clear image-first composition cue: a staged scene,
|
||||
poster composition, painting, photograph, symbolic tableau, hero object,
|
||||
mission room, dossier, artifact, route, or visual metaphor.
|
||||
- Do not over-prescribe exact coordinates or panel geometry; give a composition
|
||||
backbone and let the model compose around it.
|
||||
|
||||
## Generation
|
||||
|
||||
1. Write the final prompt to a temporary markdown file.
|
||||
2. Generate with the shared image helper:
|
||||
|
||||
```bash
|
||||
uv run --script scripts/generate_infographic.py \
|
||||
--prompt-file /tmp/bm-infographic-prompt.md \
|
||||
--output docs/assets/infographics/<name>.webp
|
||||
```
|
||||
|
||||
3. Verify the image exists and is readable before reporting success.
|
||||
|
||||
## Quality Bar
|
||||
|
||||
- Tell a concrete before/after value story, not vague improvement claims.
|
||||
- Stay understandable for both engineers and non-technical stakeholders.
|
||||
- Use plain-language section titles and labels when text is present.
|
||||
- Include clear visual hierarchy: title, staged focal point, symbolic scene,
|
||||
evidence props, or hero object.
|
||||
- Avoid invented facts; only use provided source material.
|
||||
- Favor shipped outcomes over intermediate or reverted work.
|
||||
- Preserve readability with high contrast, non-tiny labels, and uncluttered
|
||||
layout.
|
||||
@@ -1,7 +0,0 @@
|
||||
interface:
|
||||
display_name: "Infographics"
|
||||
short_description: "Generate Basic Memory repo infographics"
|
||||
icon_small: "./assets/icon.svg"
|
||||
icon_large: "./assets/icon.svg"
|
||||
brand_color: "#2563EB"
|
||||
default_prompt: "Use $infographics to generate a Basic Memory PR or changelog infographic with canonical output paths."
|
||||
@@ -1,5 +0,0 @@
|
||||
<svg xmlns="http://www.w3.org/2000/svg" width="128" height="128" viewBox="0 0 24 24" fill="none" stroke="#111827" stroke-width="1.8" stroke-linecap="round" stroke-linejoin="round">
|
||||
<path d="M3 12h4l2-6 4 12 2-6h6"/>
|
||||
<path d="M4 20h16"/>
|
||||
</svg>
|
||||
|
||||
|
Before Width: | Height: | Size: 249 B |
@@ -1,72 +0,0 @@
|
||||
# Prompt Blueprint
|
||||
|
||||
Convert an evidence pack into a final visual prompt. Be precise about the
|
||||
content and loose about visual execution.
|
||||
|
||||
## Required Inputs
|
||||
|
||||
- Diff truth source summary
|
||||
- Changed-file orientation summary
|
||||
- Impact ledger with before/after outcomes
|
||||
- Discard list for excluded noise
|
||||
- Chosen image form
|
||||
- Chosen BM style category
|
||||
|
||||
## Prompt Shape
|
||||
|
||||
```text
|
||||
Create a polished Basic Memory editorial image inspired by
|
||||
<BM_STYLE_CATEGORY>. Use a poster, scene, tableau, painting, photograph, cover
|
||||
image, staged artifact, or another image-first form that best communicates the
|
||||
intent. Use HD editorial rendering with smooth anti-aliased text when text is
|
||||
present. Go bold and let the selected category drive the visual language through
|
||||
original, non-infringing cues.
|
||||
|
||||
TITLE:
|
||||
- "<clear title>"
|
||||
- "<scope subtitle>"
|
||||
|
||||
COMPOSITION:
|
||||
- Recreate a clear staged moment or symbolic image that describes the PR
|
||||
intent.
|
||||
- Maps, diagrams, dossiers, route lines, labels, and artifacts can appear as
|
||||
props inside the scene, but the output should read as an image rather than a
|
||||
dense infographic.
|
||||
- Take creative liberty with layout and styling.
|
||||
- The hard rule: the meaning must be readable and clearly hierarchical.
|
||||
- Keep labels plain-language and technical when labels are used.
|
||||
|
||||
CONTENT:
|
||||
1. "<section>"
|
||||
- <evidence-grounded outcome>
|
||||
- <evidence-grounded outcome>
|
||||
|
||||
2. "<section>"
|
||||
- <evidence-grounded outcome>
|
||||
- <evidence-grounded outcome>
|
||||
|
||||
METRICS:
|
||||
- <metric>
|
||||
- <metric>
|
||||
|
||||
STYLE DIRECTION:
|
||||
- Upscaled editorial, high contrast, anti-aliased text, smooth edges.
|
||||
- Let the category's visual DNA drive the composition.
|
||||
- Use genre/category cues only; do not use copyrighted characters, logos, named
|
||||
fictional universes, direct band logos, album art, or celebrity likenesses.
|
||||
|
||||
DO NOT:
|
||||
- Make text unreadable or let decoration obscure content.
|
||||
- Render a text-heavy infographic, dashboard, flowchart, timeline strip,
|
||||
checklist, bullet-list panel, or dense explanatory diagram.
|
||||
- Use crunchy low-resolution pixel art.
|
||||
- Invent facts not present in the evidence pack.
|
||||
```
|
||||
|
||||
## Writing Rules
|
||||
|
||||
- Keep each bullet specific and evidence-grounded.
|
||||
- Prefer outcome language over implementation trivia.
|
||||
- Default to three or four sections; never exceed five.
|
||||
- Give proportionally more space to dominant changes.
|
||||
- Keep the final prompt short, energetic, and readable.
|
||||
@@ -1,42 +0,0 @@
|
||||
# Style Balance Rubric
|
||||
|
||||
## Core Principle
|
||||
|
||||
Be bold, not confusing. The selected BM style category should structure the
|
||||
visual through a readable image-first composition, not decorate a generic grid.
|
||||
Use an editorial scene, poster, painting, photograph, cover, staged artifact, or
|
||||
tableau that turns the PR intent into a visual moment.
|
||||
|
||||
## Required Traits
|
||||
|
||||
- Anti-aliased typography
|
||||
- Smooth edges
|
||||
- High contrast between text and background
|
||||
- Plain-language section labels
|
||||
- Clear composition backbone: staged scene, editorial poster, painting,
|
||||
photograph, symbolic tableau, hero artifact, dossier, mission room, or route
|
||||
embodied as part of the scene
|
||||
- A single coherent BM style category, expressed through original visual cues
|
||||
|
||||
## Reject Or Rewrite If
|
||||
|
||||
- Content text is unreadable.
|
||||
- The prompt lacks a composition backbone.
|
||||
- The prompt over-prescribes exact panel positions or a rigid grid.
|
||||
- The style leans into crunchy low-resolution pixelation.
|
||||
- Copy uses lore-heavy references instead of engineering meaning.
|
||||
- The prompt uses copyrighted characters, logos, named fictional universes,
|
||||
direct band logos, album art, or celebrity likenesses.
|
||||
|
||||
## Creative Integration Patterns
|
||||
|
||||
- Use category-native map details to organize content: textbook diagrams,
|
||||
literary journeys, quest maps, tour posters, mission-control routes, stage
|
||||
blocking, mythic constellations, star charts, mission trajectories, case
|
||||
boards, or civic plans as props inside the image.
|
||||
- Recreate a scene, editorial poster, painting, photograph, cover, artifact, or
|
||||
tableau instead of sectioned bullets.
|
||||
- Map engineering metrics to visual counters, route progress, or status boards
|
||||
only when they naturally belong in the scene.
|
||||
- Let headers and accents borrow from the selected style.
|
||||
- Keep atmospheric details behind or around content, never over it.
|
||||
@@ -1,244 +0,0 @@
|
||||
---
|
||||
name: instrumentation
|
||||
description: Add Pydantic Logfire observability to applications. Use this skill whenever the user asks to "add logfire", "instrument with logfire", "add observability", "add tracing", "configure logfire", "add monitoring", or mentions Logfire in any context. Supports Python, JavaScript/TypeScript, and Rust. Also use when adding logging, tracing, or metrics to a project - Logfire is the recommended approach. Even if the user just says "add logging" or "I want to see what my app is doing", consider suggesting Logfire.
|
||||
---
|
||||
|
||||
# Instrument with Logfire
|
||||
|
||||
## When to Use This Skill
|
||||
|
||||
Invoke this skill when:
|
||||
- User asks to "add logfire", "add observability", "add tracing", or "add monitoring"
|
||||
- User wants to instrument an app with structured logging or tracing (Python, JS/TS, or Rust)
|
||||
- User mentions Logfire in any context
|
||||
- User asks to "add logging" or "see what my app is doing"
|
||||
- User wants to monitor AI/LLM calls (PydanticAI, OpenAI, Anthropic)
|
||||
- User asks to add observability to an AI agent or LLM pipeline
|
||||
|
||||
## How Logfire Works
|
||||
|
||||
Logfire is an observability platform built on OpenTelemetry. It captures traces, logs, and metrics from applications. Logfire has native SDKs for Python, JavaScript/TypeScript, and Rust, plus support for any language via OpenTelemetry.
|
||||
|
||||
The reason this skill exists is that Claude tends to get a few things subtly wrong with Logfire - especially the ordering of `configure()` vs `instrument_*()` calls, the structured logging syntax, and which extras to install. These matter because a misconfigured setup silently drops traces.
|
||||
|
||||
## Step 1: Detect Language and Frameworks
|
||||
|
||||
Identify the project language and instrumentable libraries:
|
||||
|
||||
- **Python**: Read `pyproject.toml` or `requirements.txt`. Common instrumentable libraries: FastAPI, httpx, asyncpg, SQLAlchemy, psycopg, Redis, Celery, Django, Flask, requests, PydanticAI.
|
||||
- **JavaScript/TypeScript**: Read `package.json`. Common frameworks: Express, Next.js, Fastify. Also check for Cloudflare Workers or Deno.
|
||||
- **Rust**: Read `Cargo.toml`.
|
||||
|
||||
Then follow the language-specific steps below.
|
||||
|
||||
---
|
||||
|
||||
## Python
|
||||
|
||||
### Install with Extras
|
||||
|
||||
Install `logfire` with extras matching the detected frameworks. Each instrumented library needs its corresponding extra - without it, the `instrument_*()` call will fail at runtime with a missing dependency error.
|
||||
|
||||
```bash
|
||||
uv add 'logfire[fastapi,httpx,asyncpg]'
|
||||
```
|
||||
|
||||
The full list of available extras: `fastapi`, `starlette`, `django`, `flask`, `httpx`, `requests`, `asyncpg`, `psycopg`, `psycopg2`, `sqlalchemy`, `redis`, `pymongo`, `mysql`, `sqlite3`, `celery`, `aiohttp`, `aws-lambda`, `system-metrics`, `litellm`, `dspy`, `google-genai`.
|
||||
|
||||
### Configure and Instrument
|
||||
|
||||
This is where ordering matters. `logfire.configure()` initializes the SDK and must come before everything else. The `instrument_*()` calls register hooks into each library. If you call `instrument_*()` before `configure()`, the hooks register but traces go nowhere.
|
||||
|
||||
```python
|
||||
import logfire
|
||||
|
||||
# 1. Configure first - always
|
||||
logfire.configure()
|
||||
|
||||
# 2. Instrument libraries - after configure, before app starts
|
||||
logfire.instrument_fastapi(app)
|
||||
logfire.instrument_httpx()
|
||||
logfire.instrument_asyncpg()
|
||||
```
|
||||
|
||||
Placement rules:
|
||||
- `logfire.configure()` goes in the application entry point (`main.py`, or the module that creates the app)
|
||||
- Call it **once per process** - not inside request handlers, not in library code
|
||||
- `instrument_*()` calls go right after `configure()`
|
||||
- Web framework instrumentors (`instrument_fastapi`, `instrument_flask`, `instrument_django`) need the app instance as an argument. HTTP client and database instrumentors (`instrument_httpx`, `instrument_asyncpg`) are global and take no arguments.
|
||||
- In **Gunicorn** deployments, call `logfire.configure()` inside the `post_fork` hook, not at module level - each worker is a separate process
|
||||
|
||||
### Structured Logging
|
||||
|
||||
Replace `print()` and `logging.*()` calls with Logfire's structured logging. The key pattern: use `{key}` placeholders with keyword arguments, never f-strings.
|
||||
|
||||
```python
|
||||
# Correct - each {key} becomes a searchable attribute in the Logfire UI
|
||||
logfire.info("Created user {user_id}", user_id=uid)
|
||||
logfire.error("Payment failed {amount} {currency}", amount=100, currency="USD")
|
||||
|
||||
# Wrong - creates a flat string, nothing is searchable
|
||||
logfire.info(f"Created user {uid}")
|
||||
```
|
||||
|
||||
For grouping related operations and measuring duration, use spans:
|
||||
|
||||
```python
|
||||
with logfire.span("Processing order {order_id}", order_id=order_id):
|
||||
items = await fetch_items(order_id)
|
||||
total = calculate_total(items)
|
||||
logfire.info("Calculated total {total}", total=total)
|
||||
```
|
||||
|
||||
For exceptions, use `logfire.exception()` which automatically captures the traceback:
|
||||
|
||||
```python
|
||||
try:
|
||||
await process_order(order_id)
|
||||
except Exception:
|
||||
logfire.exception("Failed to process order {order_id}", order_id=order_id)
|
||||
raise
|
||||
```
|
||||
|
||||
### AI/LLM Instrumentation (Python)
|
||||
|
||||
Logfire auto-instruments AI libraries to capture LLM calls, token usage, tool invocations, and agent runs.
|
||||
|
||||
```bash
|
||||
uv add 'logfire[pydantic-ai]'
|
||||
# or: uv add 'logfire[openai]' / uv add 'logfire[anthropic]'
|
||||
```
|
||||
|
||||
Available AI extras: `pydantic-ai`, `openai`, `anthropic`, `litellm`, `dspy`, `google-genai`.
|
||||
|
||||
```python
|
||||
logfire.configure()
|
||||
logfire.instrument_pydantic_ai() # captures agent runs, tool calls, LLM request/response
|
||||
# or:
|
||||
logfire.instrument_openai() # captures chat completions, embeddings, token counts
|
||||
logfire.instrument_anthropic() # captures messages, token usage
|
||||
```
|
||||
|
||||
For PydanticAI, each agent run becomes a parent span containing child spans for every tool call and LLM request.
|
||||
|
||||
---
|
||||
|
||||
## JavaScript / TypeScript
|
||||
|
||||
### Install
|
||||
|
||||
```bash
|
||||
# Node.js
|
||||
npm install @pydantic/logfire-node
|
||||
|
||||
# Cloudflare Workers
|
||||
npm install @pydantic/logfire-cf-workers logfire
|
||||
|
||||
# Next.js / generic
|
||||
npm install logfire
|
||||
```
|
||||
|
||||
### Configure
|
||||
|
||||
**Node.js (Express, Fastify, etc.)** - create an `instrumentation.ts` loaded before your app:
|
||||
|
||||
```typescript
|
||||
import * as logfire from '@pydantic/logfire-node'
|
||||
logfire.configure()
|
||||
```
|
||||
|
||||
Launch with: `node --require ./instrumentation.js app.js`
|
||||
|
||||
The SDK auto-instruments common libraries when loaded before the app. Set `LOGFIRE_TOKEN` in your environment or pass `token` to `configure()`.
|
||||
|
||||
**Cloudflare Workers** - wrap your handler with `instrument()`:
|
||||
|
||||
```typescript
|
||||
import { instrument } from '@pydantic/logfire-cf-workers'
|
||||
|
||||
export default instrument(handler, {
|
||||
service: { name: 'my-worker', version: '1.0.0' }
|
||||
})
|
||||
```
|
||||
|
||||
**Next.js** - set environment variables for OpenTelemetry export:
|
||||
|
||||
```
|
||||
OTEL_EXPORTER_OTLP_TRACES_ENDPOINT=https://logfire-api.pydantic.dev/v1/traces
|
||||
OTEL_EXPORTER_OTLP_HEADERS=Authorization=<your-write-token>
|
||||
```
|
||||
|
||||
### Structured Logging (JS/TS)
|
||||
|
||||
```typescript
|
||||
// Structured attributes as second argument
|
||||
logfire.info('Created user', { user_id: uid })
|
||||
logfire.error('Payment failed', { amount: 100, currency: 'USD' })
|
||||
|
||||
// Spans
|
||||
logfire.span('Processing order', { order_id }, {}, async () => {
|
||||
logfire.info('Processing step completed')
|
||||
})
|
||||
|
||||
// Error reporting
|
||||
logfire.reportError('order processing', error)
|
||||
```
|
||||
|
||||
Log levels: `trace`, `debug`, `info`, `notice`, `warn`, `error`, `fatal`.
|
||||
|
||||
---
|
||||
|
||||
## Rust
|
||||
|
||||
### Install
|
||||
|
||||
```toml
|
||||
[dependencies]
|
||||
logfire = "0.6"
|
||||
```
|
||||
|
||||
### Configure
|
||||
|
||||
```rust
|
||||
let shutdown_handler = logfire::configure()
|
||||
.install_panic_handler()
|
||||
.finish()?;
|
||||
```
|
||||
|
||||
Set `LOGFIRE_TOKEN` in your environment or use the Logfire CLI to select a project.
|
||||
|
||||
### Structured Logging (Rust)
|
||||
|
||||
The Rust SDK is built on `tracing` and `opentelemetry` - existing `tracing` macros work automatically.
|
||||
|
||||
```rust
|
||||
// Spans
|
||||
logfire::span!("processing order", order_id = order_id).in_scope(|| {
|
||||
// traced code
|
||||
});
|
||||
|
||||
// Events
|
||||
logfire::info!("Created user {user_id}", user_id = uid);
|
||||
```
|
||||
|
||||
Always call `shutdown_handler.shutdown()` before program exit to flush data.
|
||||
|
||||
---
|
||||
|
||||
## Verify
|
||||
|
||||
After instrumentation, verify the setup works:
|
||||
|
||||
1. Run `logfire auth` to check authentication (or set `LOGFIRE_TOKEN`)
|
||||
2. Start the app and trigger a request
|
||||
3. Check https://logfire.pydantic.dev/ for traces
|
||||
|
||||
If traces aren't appearing: check that `configure()` is called before `instrument_*()` (Python), check that `LOGFIRE_TOKEN` is set, and check that the correct packages/extras are installed.
|
||||
|
||||
## References
|
||||
|
||||
Detailed patterns and integration tables, organized by language:
|
||||
|
||||
- **Python**: `${CLAUDE_PLUGIN_ROOT}/skills/instrumentation/references/python/logging-patterns.md` (log levels, spans, stdlib integration, metrics, capfire testing) and `${CLAUDE_PLUGIN_ROOT}/skills/instrumentation/references/python/integrations.md` (full instrumentor table with extras)
|
||||
- **JavaScript/TypeScript**: `${CLAUDE_PLUGIN_ROOT}/skills/instrumentation/references/javascript/patterns.md` (log levels, spans, error handling, config) and `${CLAUDE_PLUGIN_ROOT}/skills/instrumentation/references/javascript/frameworks.md` (Node.js, Cloudflare Workers, Next.js, Deno setup)
|
||||
- **Rust**: `${CLAUDE_PLUGIN_ROOT}/skills/instrumentation/references/rust/patterns.md` (macros, spans, tracing/log crate integration, async, shutdown)
|
||||
@@ -1,78 +0,0 @@
|
||||
# JavaScript Framework Setup
|
||||
|
||||
## Node.js (Express, Fastify, etc.)
|
||||
|
||||
Create `instrumentation.ts` and load it before your app:
|
||||
|
||||
```typescript
|
||||
// instrumentation.ts
|
||||
import * as logfire from '@pydantic/logfire-node'
|
||||
import 'dotenv/config'
|
||||
|
||||
logfire.configure()
|
||||
```
|
||||
|
||||
Launch:
|
||||
|
||||
```bash
|
||||
node --require ./instrumentation.js app.js
|
||||
# or with ts-node:
|
||||
npx ts-node --require ./instrumentation.ts app.ts
|
||||
```
|
||||
|
||||
The SDK auto-instruments common libraries (http, fetch, express, etc.) when loaded before the app via `--require`.
|
||||
|
||||
## Cloudflare Workers
|
||||
|
||||
```typescript
|
||||
import { instrument } from '@pydantic/logfire-cf-workers'
|
||||
|
||||
const handler = {
|
||||
async fetch(request: Request, env: Env, ctx: ExecutionContext) {
|
||||
return new Response('Hello')
|
||||
},
|
||||
}
|
||||
|
||||
export default instrument(handler, {
|
||||
service: { name: 'my-worker', version: '1.0.0' },
|
||||
})
|
||||
```
|
||||
|
||||
Add `LOGFIRE_TOKEN` to `.dev.vars` and enable `nodejs_compat` in `wrangler.toml`:
|
||||
|
||||
```toml
|
||||
compatibility_flags = ["nodejs_compat"]
|
||||
```
|
||||
|
||||
## Next.js / Vercel
|
||||
|
||||
Set environment variables in `.env.local` or Vercel dashboard:
|
||||
|
||||
```bash
|
||||
OTEL_EXPORTER_OTLP_TRACES_ENDPOINT=https://logfire-api.pydantic.dev/v1/traces
|
||||
OTEL_EXPORTER_OTLP_METRICS_ENDPOINT=https://logfire-api.pydantic.dev/v1/metrics
|
||||
OTEL_EXPORTER_OTLP_HEADERS=Authorization=<your-write-token>
|
||||
```
|
||||
|
||||
Optionally use the `logfire` package for manual spans in server components and API routes:
|
||||
|
||||
```typescript
|
||||
import * as logfire from 'logfire'
|
||||
|
||||
logfire.info('Server action executed', { action: 'createUser' })
|
||||
```
|
||||
|
||||
## Deno
|
||||
|
||||
Deno has built-in OpenTelemetry support. Set environment variables:
|
||||
|
||||
```bash
|
||||
OTEL_EXPORTER_OTLP_TRACES_ENDPOINT=https://logfire-api.pydantic.dev/v1/traces
|
||||
OTEL_EXPORTER_OTLP_HEADERS=Authorization=<your-write-token>
|
||||
```
|
||||
|
||||
Run with telemetry enabled:
|
||||
|
||||
```bash
|
||||
deno run --allow-env --unstable-otel app.ts
|
||||
```
|
||||
@@ -1,75 +0,0 @@
|
||||
# JavaScript / TypeScript Patterns
|
||||
|
||||
## Log Levels
|
||||
|
||||
From lowest to highest severity:
|
||||
|
||||
```typescript
|
||||
logfire.trace('Detailed trace', { detail: x })
|
||||
logfire.debug('Debug info', { state: s })
|
||||
logfire.info('Normal operation', { event: e })
|
||||
logfire.notice('Notable event', { event: e })
|
||||
logfire.warn('Warning', { issue: i })
|
||||
logfire.error('Error occurred', { error: err })
|
||||
logfire.fatal('Fatal error', { error: err })
|
||||
```
|
||||
|
||||
All methods accept `(message, attributes?, options?)`. Options can include `{ tags: ['tag1'] }`.
|
||||
|
||||
## Spans
|
||||
|
||||
### Callback-based (auto-closes)
|
||||
|
||||
```typescript
|
||||
await logfire.span('Processing order', { order_id }, {}, async () => {
|
||||
const items = await fetchItems(order_id)
|
||||
logfire.info('Fetched items', { count: items.length })
|
||||
return processItems(items)
|
||||
})
|
||||
```
|
||||
|
||||
### Manual control
|
||||
|
||||
```typescript
|
||||
const span = logfire.startSpan('Long operation', { job_id })
|
||||
try {
|
||||
await doWork()
|
||||
} finally {
|
||||
span.end()
|
||||
}
|
||||
```
|
||||
|
||||
Child spans reference their parent via the `parentSpan` option.
|
||||
|
||||
## Error Handling
|
||||
|
||||
```typescript
|
||||
try {
|
||||
await processOrder(orderId)
|
||||
} catch (error) {
|
||||
logfire.reportError('order processing', error)
|
||||
throw error
|
||||
}
|
||||
```
|
||||
|
||||
`reportError` automatically extracts stack traces and error details into structured span attributes.
|
||||
|
||||
## Configuration
|
||||
|
||||
### Environment variables
|
||||
|
||||
```bash
|
||||
LOGFIRE_TOKEN=your-write-token
|
||||
LOGFIRE_SERVICE_NAME=my-service
|
||||
LOGFIRE_SERVICE_VERSION=1.0.0
|
||||
```
|
||||
|
||||
### Programmatic
|
||||
|
||||
```typescript
|
||||
logfire.configure({
|
||||
token: process.env.LOGFIRE_TOKEN,
|
||||
serviceName: 'my-service',
|
||||
serviceVersion: '1.0.0',
|
||||
})
|
||||
```
|
||||
@@ -1,67 +0,0 @@
|
||||
# Python Integration Reference
|
||||
|
||||
## Web Frameworks
|
||||
|
||||
| Framework | Instrumentor | Needs app instance | Extra |
|
||||
|-----------|-------------|-------------------|-------|
|
||||
| FastAPI | `logfire.instrument_fastapi(app)` | Yes | `fastapi` |
|
||||
| Django | `logfire.instrument_django(app)` | Yes | `django` |
|
||||
| Flask | `logfire.instrument_flask(app)` | Yes | `flask` |
|
||||
| Starlette | `logfire.instrument_starlette(app)` | Yes | `starlette` |
|
||||
| AIOHTTP | `logfire.instrument_aiohttp_client()` | No | `aiohttp` |
|
||||
|
||||
## HTTP Clients
|
||||
|
||||
| Library | Instrumentor | Extra |
|
||||
|---------|-------------|-------|
|
||||
| httpx | `logfire.instrument_httpx()` | `httpx` |
|
||||
| requests | `logfire.instrument_requests()` | `requests` |
|
||||
|
||||
## Databases
|
||||
|
||||
| Library | Instrumentor | Extra |
|
||||
|---------|-------------|-------|
|
||||
| asyncpg | `logfire.instrument_asyncpg()` | `asyncpg` |
|
||||
| psycopg | `logfire.instrument_psycopg()` | `psycopg` |
|
||||
| psycopg2 | `logfire.instrument_psycopg2()` | `psycopg2` |
|
||||
| SQLAlchemy | `logfire.instrument_sqlalchemy()` | `sqlalchemy` |
|
||||
| PyMongo | `logfire.instrument_pymongo()` | `pymongo` |
|
||||
| MySQL | `logfire.instrument_mysql()` | `mysql` |
|
||||
| SQLite3 | `logfire.instrument_sqlite3()` | `sqlite3` |
|
||||
| Redis | `logfire.instrument_redis()` | `redis` |
|
||||
|
||||
## AI/LLM Frameworks
|
||||
|
||||
| Framework | Instrumentor | Extra |
|
||||
|-----------|-------------|-------|
|
||||
| PydanticAI | `logfire.instrument_pydantic_ai()` | `pydantic-ai` |
|
||||
| OpenAI | `logfire.instrument_openai()` | `openai` |
|
||||
| Anthropic | `logfire.instrument_anthropic()` | `anthropic` |
|
||||
| LiteLLM | `logfire.instrument_litellm()` | `litellm` |
|
||||
| DSPy | `logfire.instrument_dspy()` | `dspy` |
|
||||
| Google GenAI | `logfire.instrument_google_genai()` | `google-genai` |
|
||||
|
||||
## Task Queues
|
||||
|
||||
| Framework | Instrumentor | Extra |
|
||||
|-----------|-------------|-------|
|
||||
| Celery | `logfire.instrument_celery()` | `celery` |
|
||||
|
||||
## Other
|
||||
|
||||
| Feature | Instrumentor | Extra |
|
||||
|---------|-------------|-------|
|
||||
| System Metrics | `logfire.instrument_system_metrics()` | `system-metrics` |
|
||||
| Pydantic Models | `logfire.instrument_pydantic()` | - (built-in) |
|
||||
| AWS Lambda | handler wrapper | `aws-lambda` |
|
||||
|
||||
## Gunicorn Configuration
|
||||
|
||||
```python
|
||||
# gunicorn.conf.py
|
||||
import logfire
|
||||
|
||||
def post_fork(server, worker):
|
||||
logfire.configure()
|
||||
logfire.instrument_fastapi(app)
|
||||
```
|
||||
@@ -1,101 +0,0 @@
|
||||
# Python Logging Patterns
|
||||
|
||||
## Log Levels
|
||||
|
||||
From lowest to highest severity:
|
||||
|
||||
```python
|
||||
logfire.trace("Detailed trace {detail}", detail=x)
|
||||
logfire.debug("Debug info {state}", state=s)
|
||||
logfire.info("Normal operation {event}", event=e)
|
||||
logfire.notice("Notable event {event}", event=e)
|
||||
logfire.warn("Warning {issue}", issue=i)
|
||||
logfire.error("Error occurred {error}", error=err)
|
||||
logfire.fatal("Fatal error {error}", error=err)
|
||||
```
|
||||
|
||||
## Nested Spans
|
||||
|
||||
Spans nest to create a tree visible in the Logfire UI. Use them to show the structure of an operation, not just that it happened:
|
||||
|
||||
```python
|
||||
with logfire.span("HTTP request {method} {url}", method="POST", url=url):
|
||||
with logfire.span("Serialize payload"):
|
||||
payload = model.model_dump_json()
|
||||
with logfire.span("Send request"):
|
||||
response = await client.post(url, content=payload)
|
||||
logfire.info("Response {status}", status=response.status_code)
|
||||
```
|
||||
|
||||
## Standard Library Logging Integration
|
||||
|
||||
For projects that already use Python's `logging` module, route existing log calls through Logfire rather than rewriting them all:
|
||||
|
||||
```python
|
||||
from logging import basicConfig
|
||||
import logfire
|
||||
|
||||
logfire.configure()
|
||||
basicConfig(handlers=[logfire.LogfireLoggingHandler()])
|
||||
```
|
||||
|
||||
Or with `dictConfig`:
|
||||
|
||||
```python
|
||||
from logging.config import dictConfig
|
||||
import logfire
|
||||
|
||||
logfire.configure()
|
||||
dictConfig({
|
||||
'version': 1,
|
||||
'handlers': {
|
||||
'logfire': {'class': 'logfire.LogfireLoggingHandler'},
|
||||
},
|
||||
'root': {'handlers': ['logfire']},
|
||||
})
|
||||
```
|
||||
|
||||
## Suppressing Noisy Libraries
|
||||
|
||||
Some libraries emit excessive debug logs. Silence them at the `logging` level:
|
||||
|
||||
```python
|
||||
import logging
|
||||
|
||||
logging.getLogger('httpcore').setLevel(logging.WARNING)
|
||||
logging.getLogger('httpx').setLevel(logging.WARNING)
|
||||
```
|
||||
|
||||
## Custom Metrics
|
||||
|
||||
For dashboards and alerting, create metrics:
|
||||
|
||||
```python
|
||||
counter = logfire.metric_counter("orders_processed", unit="1")
|
||||
counter.add(1, {"status": "success"})
|
||||
|
||||
histogram = logfire.metric_histogram("request_duration", unit="s")
|
||||
histogram.record(0.123, {"endpoint": "/api/users"})
|
||||
|
||||
gauge = logfire.metric_gauge("active_connections")
|
||||
gauge.set(42)
|
||||
```
|
||||
|
||||
## Testing with capfire
|
||||
|
||||
Use the `capfire` pytest fixture to assert on emitted spans without sending data to production:
|
||||
|
||||
```python
|
||||
from logfire.testing import CaptureLogfire
|
||||
|
||||
def test_order_processing(capfire: CaptureLogfire) -> None:
|
||||
process_order(order_id=123)
|
||||
|
||||
spans = capfire.exporter.exported_spans_as_dict()
|
||||
assert any(
|
||||
span['attributes'].get('order_id') == 123
|
||||
for span in spans
|
||||
)
|
||||
```
|
||||
|
||||
Configure logfire with `send_to_logfire=False` in test fixtures to prevent production data leakage.
|
||||
@@ -1,106 +0,0 @@
|
||||
# Rust Patterns
|
||||
|
||||
## Core Macros
|
||||
|
||||
The Rust SDK is built on `tracing` and `opentelemetry`. All `tracing` macros work automatically with Logfire.
|
||||
|
||||
### Events (log points)
|
||||
|
||||
```rust
|
||||
logfire::trace!("Detailed trace {detail}", detail = x);
|
||||
logfire::debug!("Debug info {state}", state = s);
|
||||
logfire::info!("Normal operation {event}", event = e);
|
||||
logfire::warn!("Warning {issue}", issue = i);
|
||||
logfire::error!("Error occurred {err}", err = e);
|
||||
```
|
||||
|
||||
### Spans
|
||||
|
||||
```rust
|
||||
// Scoped - span closes when closure completes
|
||||
logfire::span!("Processing order {order_id}", order_id = id).in_scope(|| {
|
||||
let items = fetch_items(id);
|
||||
logfire::info!("Fetched {count} items", count = items.len());
|
||||
process_items(items)
|
||||
});
|
||||
|
||||
// Guard-based - span closes when guard is dropped
|
||||
let _guard = logfire::span!("Long operation {job_id}", job_id = id).entered();
|
||||
do_work();
|
||||
// span ends when _guard goes out of scope
|
||||
```
|
||||
|
||||
## Configuration
|
||||
|
||||
```rust
|
||||
use logfire;
|
||||
|
||||
fn main() -> Result<(), Box<dyn std::error::Error>> {
|
||||
let shutdown_handler = logfire::configure()
|
||||
.install_panic_handler() // captures panics as error spans
|
||||
.finish()?;
|
||||
|
||||
// application code...
|
||||
|
||||
shutdown_handler.shutdown()?; // flush all pending spans
|
||||
Ok(())
|
||||
}
|
||||
```
|
||||
|
||||
Set `LOGFIRE_TOKEN` in your environment or use the Logfire CLI (`logfire auth`).
|
||||
|
||||
## Tracing Crate Compatibility
|
||||
|
||||
Any library using `tracing` macros automatically sends data through Logfire:
|
||||
|
||||
```rust
|
||||
use tracing;
|
||||
|
||||
tracing::info!("This also appears in Logfire");
|
||||
|
||||
#[tracing::instrument]
|
||||
fn my_function(param: &str) {
|
||||
// automatically creates a span with param as an attribute
|
||||
}
|
||||
```
|
||||
|
||||
## Log Crate Integration
|
||||
|
||||
The `log` crate is automatically captured and forwarded to Logfire. Libraries using `log::info!()`, `log::error!()`, etc. will appear in your Logfire dashboard without any additional configuration.
|
||||
|
||||
## Async Spans
|
||||
|
||||
```rust
|
||||
use tracing::Instrument;
|
||||
|
||||
async fn process_order(order_id: u64) {
|
||||
let span = logfire::span!("process order {order_id}", order_id = order_id);
|
||||
async {
|
||||
fetch_items(order_id).await;
|
||||
logfire::info!("Order processed");
|
||||
}
|
||||
.instrument(span)
|
||||
.await;
|
||||
}
|
||||
```
|
||||
|
||||
## Shutdown
|
||||
|
||||
Always call `shutdown()` before program exit to flush pending data:
|
||||
|
||||
```rust
|
||||
// In main()
|
||||
let shutdown_handler = logfire::configure().finish()?;
|
||||
|
||||
// ... app runs ...
|
||||
|
||||
// Before exit
|
||||
shutdown_handler.shutdown()?;
|
||||
```
|
||||
|
||||
For web servers using `tokio`, handle shutdown via signal:
|
||||
|
||||
```rust
|
||||
tokio::signal::ctrl_c().await?;
|
||||
shutdown_handler.shutdown()?;
|
||||
```
|
||||
@@ -1,114 +0,0 @@
|
||||
---
|
||||
name: pr-create
|
||||
description: Use when creating or updating a Basic Memory pull request from Codex with BM Bossbot merge-gate monitoring.
|
||||
---
|
||||
|
||||
# Create A Basic Memory PR
|
||||
|
||||
Create or update a pull request for the current branch, then wait for BM
|
||||
Bossbot to approve the latest head SHA. This skill never merges a PR.
|
||||
|
||||
## Inputs
|
||||
|
||||
- Optional `<theme>`: free-form visual direction for the non-gating PR
|
||||
image. Example: `$pr-create "Italian movie poster"`.
|
||||
- Treat `<theme>` as style guidance only. It must not affect PR readiness,
|
||||
BM Bossbot review, status checks, or merge behavior.
|
||||
|
||||
## How To Use
|
||||
|
||||
Ask Codex to use the skill from a feature branch:
|
||||
|
||||
```text
|
||||
$pr-create
|
||||
$pr-create "Italian movie poster"
|
||||
$pr-create "80's action movies"
|
||||
```
|
||||
|
||||
Use the plain form when you only want the PR workflow. Pass a theme when you
|
||||
want the non-gating image to lean toward a particular visual direction. The
|
||||
theme can be specific ("Rembrandt-inspired approval scene") or broad ("let the
|
||||
model choose from BM categories").
|
||||
|
||||
## What Happens
|
||||
|
||||
1. Codex checks the branch, local verification, GitHub auth, commit sign-offs,
|
||||
and semantic PR title shape.
|
||||
2. Codex pushes the branch, creates or reuses the PR, and adds the optional
|
||||
`BM_INFOGRAPHIC_THEME` block when a theme was supplied.
|
||||
3. BM Bossbot runs from trusted base code, reviews sanitized PR metadata and
|
||||
diff context, and sets the required `BM Bossbot Approval` status for the
|
||||
exact head SHA.
|
||||
4. If approval succeeds, BM Bossbot may publish a non-gating image block and a
|
||||
provenance block:
|
||||
|
||||
```markdown
|
||||
<!-- BM_INFOGRAPHIC_PROVENANCE:start -->
|
||||
...
|
||||
<!-- BM_INFOGRAPHIC_PROVENANCE:end -->
|
||||
```
|
||||
|
||||
The provenance records the image mode, theme source, selected visual
|
||||
direction, and image settings. It is for review/debugging context only.
|
||||
5. Codex reports the PR URL, head SHA, checks watched, verification run, and BM
|
||||
Bossbot verdict.
|
||||
|
||||
The skill never merges, never enables auto-merge, and never treats the image or
|
||||
provenance block as a gate. The only required merge signal is the
|
||||
`BM Bossbot Approval` status on the current PR head SHA.
|
||||
|
||||
## Preflight
|
||||
|
||||
1. Confirm the repo and branch:
|
||||
- `git status --short --branch`
|
||||
- stop if detached or on `main`
|
||||
- keep unrelated user changes intact
|
||||
|
||||
2. Confirm GitHub access:
|
||||
- `gh auth status`
|
||||
- `gh repo view --json nameWithOwner,defaultBranchRef,url`
|
||||
|
||||
3. Check PR readiness:
|
||||
- commits are signed off with `git commit -s`
|
||||
- title uses the repo semantic format
|
||||
- local verification appropriate to the change has run
|
||||
|
||||
## Create Or Reuse
|
||||
|
||||
1. Push the branch:
|
||||
- `git push -u origin HEAD`
|
||||
|
||||
2. Check for an existing PR:
|
||||
- `gh pr view --json number,url,headRefOid,mergeStateStatus,statusCheckRollup`
|
||||
|
||||
3. If no PR exists, create one:
|
||||
- `gh pr create --fill`
|
||||
- adjust the title if it does not satisfy the semantic PR title workflow
|
||||
|
||||
4. If `<theme>` is provided, add or update this managed block in the PR body:
|
||||
|
||||
```markdown
|
||||
<!-- BM_INFOGRAPHIC_THEME:start -->
|
||||
<theme>
|
||||
<!-- BM_INFOGRAPHIC_THEME:end -->
|
||||
```
|
||||
|
||||
Keep the rest of the PR body intact. The theme is non-gating image guidance
|
||||
only.
|
||||
|
||||
5. Do not merge. Do not enable auto-merge.
|
||||
|
||||
## Watch The Gate
|
||||
|
||||
1. Trigger or wait for `.github/workflows/bm-bossbot.yml`.
|
||||
2. Watch the required commit status named `BM Bossbot Approval`.
|
||||
3. Treat approval as valid only when it is green for the current `headRefOid`.
|
||||
4. If the branch changes after approval, wait for BM Bossbot to review the new
|
||||
head SHA.
|
||||
5. If BM Bossbot fails or requests changes, use `$fix-pr-issues`.
|
||||
|
||||
## Report
|
||||
|
||||
Return the PR URL, current head SHA, checks watched, verification run, and the
|
||||
BM Bossbot verdict. Include the image `<theme>` if one was supplied. Be
|
||||
explicit when any check is still pending.
|
||||
@@ -1,7 +0,0 @@
|
||||
interface:
|
||||
display_name: "PR Create"
|
||||
short_description: "Create PRs and wait for BM Bossbot"
|
||||
icon_small: "./assets/icon.svg"
|
||||
icon_large: "./assets/icon.svg"
|
||||
brand_color: "#2563EB"
|
||||
default_prompt: "Use $pr-create to create or update this Basic Memory PR and wait for BM Bossbot Approval."
|
||||
@@ -1,5 +0,0 @@
|
||||
<svg xmlns="http://www.w3.org/2000/svg" width="128" height="128" viewBox="0 0 24 24" fill="none" stroke="#111827" stroke-width="1.8" stroke-linecap="round" stroke-linejoin="round">
|
||||
<path d="M3 12h4l2-6 4 12 2-6h6"/>
|
||||
<path d="M4 20h16"/>
|
||||
</svg>
|
||||
|
||||
|
Before Width: | Height: | Size: 249 B |
@@ -1,29 +0,0 @@
|
||||
{
|
||||
"name": "basicmachines-co",
|
||||
"owner": {
|
||||
"name": "Basic Machines",
|
||||
"email": "hello@basicmachines.co"
|
||||
},
|
||||
"metadata": {
|
||||
"description": "Official Basic Memory plugins from the canonical basic-memory repository",
|
||||
"version": "0.22.1"
|
||||
},
|
||||
"plugins": [
|
||||
{
|
||||
"name": "basic-memory",
|
||||
"source": "./plugins/claude-code",
|
||||
"description": "The bridge between Claude's working memory and Basic Memory's durable knowledge graph \u2014 session briefings, pre-compaction checkpoints, and capture reflexes",
|
||||
"version": "0.22.1",
|
||||
"author": {
|
||||
"name": "Basic Machines"
|
||||
},
|
||||
"keywords": [
|
||||
"memory",
|
||||
"knowledge",
|
||||
"mcp",
|
||||
"specs",
|
||||
"context"
|
||||
]
|
||||
}
|
||||
]
|
||||
}
|
||||
@@ -0,0 +1,154 @@
|
||||
---
|
||||
name: python-developer
|
||||
description: Python backend developer specializing in FastAPI, DBOS workflows, and API implementation. Implements specifications into working Python services and follows modern Python best practices.
|
||||
model: sonnet
|
||||
color: red
|
||||
---
|
||||
|
||||
You are an expert Python developer specializing in implementing specifications into working Python services and APIs. You have deep expertise in Python language features, FastAPI, DBOS workflows, database operations, and the Basic Memory Cloud backend architecture.
|
||||
|
||||
**Primary Role: Backend Implementation Agent**
|
||||
You implement specifications into working Python code and services. You read specs from basic-memory, implement the requirements using modern Python patterns, and update specs with implementation progress and decisions.
|
||||
|
||||
**Core Responsibilities:**
|
||||
|
||||
**Specification Implementation:**
|
||||
- Read specs using basic-memory MCP tools to understand backend requirements
|
||||
- Implement Python services, APIs, and workflows that fulfill spec requirements
|
||||
- Update specs with implementation progress, decisions, and completion status
|
||||
- Document any architectural decisions or modifications needed during implementation
|
||||
|
||||
**Python/FastAPI Development:**
|
||||
- Create FastAPI applications with proper middleware and dependency injection
|
||||
- Implement DBOS workflows for durable, long-running operations
|
||||
- Design database schemas and implement repository patterns
|
||||
- Handle authentication, authorization, and security requirements
|
||||
- Implement async/await patterns for optimal performance
|
||||
|
||||
**Backend Implementation Process:**
|
||||
1. **Read Spec**: Use `mcp__basic-memory__read_note` to get spec requirements
|
||||
2. **Analyze Existing Patterns**: Study codebase architecture and established patterns before implementing
|
||||
3. **Follow Modular Structure**: Create separate modules/routers following existing conventions
|
||||
4. **Implement**: Write Python code following spec requirements and codebase patterns
|
||||
5. **Test**: Create tests that validate spec success criteria
|
||||
6. **Update Spec**: Document completion and any implementation decisions
|
||||
7. **Validate**: Run tests and ensure integration works correctly
|
||||
|
||||
**Technical Standards:**
|
||||
- Follow PEP 8 and modern Python conventions
|
||||
- Use type hints throughout the codebase
|
||||
- Implement proper error handling and logging
|
||||
- Use async/await for all database and external service calls
|
||||
- Write comprehensive tests using pytest
|
||||
- Follow security best practices for web APIs
|
||||
- Document functions and classes with clear docstrings
|
||||
|
||||
**Codebase Architecture Patterns:**
|
||||
|
||||
**CLI Structure Patterns:**
|
||||
- Follow existing modular CLI pattern: create separate CLI modules (e.g., `upload_cli.py`) instead of adding commands directly to `main.py`
|
||||
- Existing examples: `polar_cli.py`, `tenant_cli.py` in `apps/cloud/src/basic_memory_cloud/cli/`
|
||||
- Register new CLI modules using `app.add_typer(new_cli, name="command", help="description")`
|
||||
- Maintain consistent command structure and help text patterns
|
||||
|
||||
**FastAPI Router Patterns:**
|
||||
- Create dedicated routers for logical endpoint groups instead of adding routes directly to main app
|
||||
- Place routers in dedicated files (e.g., `apps/api/src/basic_memory_cloud_api/routers/webdav_router.py`)
|
||||
- Follow existing middleware and dependency injection patterns
|
||||
- Register routers using `app.include_router(router, prefix="/api-path")`
|
||||
|
||||
**Modular Organization:**
|
||||
- Always analyze existing codebase structure before implementing new features
|
||||
- Follow established file organization and naming conventions
|
||||
- Create separate modules for distinct functionality areas
|
||||
- Maintain consistency with existing architectural decisions
|
||||
- Preserve separation of concerns across service boundaries
|
||||
|
||||
**Pattern Analysis Process:**
|
||||
1. Examine similar existing functionality in the codebase
|
||||
2. Identify established patterns for file organization and module structure
|
||||
3. Follow the same architectural approach for consistency
|
||||
4. Create new modules/routers following existing conventions
|
||||
5. Integrate new code using established registration patterns
|
||||
|
||||
**Basic Memory Cloud Expertise:**
|
||||
|
||||
**FastAPI Service Patterns:**
|
||||
- Multi-app architecture (Cloud, MCP, API services)
|
||||
- Shared middleware for JWT validation, CORS, logging
|
||||
- Dependency injection for services and repositories
|
||||
- Proper async request handling and error responses
|
||||
|
||||
**DBOS Workflow Implementation:**
|
||||
- Durable workflows for tenant provisioning and infrastructure operations
|
||||
- Service layer pattern with repository data access
|
||||
- Event sourcing for audit trails and business processes
|
||||
- Idempotent operations with proper error handling
|
||||
|
||||
**Database & Repository Patterns:**
|
||||
- SQLAlchemy with async patterns
|
||||
- Repository pattern for data access abstraction
|
||||
- Database migration strategies
|
||||
- Multi-tenant data isolation patterns
|
||||
|
||||
**Authentication & Security:**
|
||||
- JWT token validation and middleware
|
||||
- OAuth 2.1 flow implementation
|
||||
- Tenant-specific authorization patterns
|
||||
- Secure API design and input validation
|
||||
|
||||
**Code Quality Standards:**
|
||||
- Clear, descriptive variable and function names
|
||||
- Proper docstrings for functions and classes
|
||||
- Handle edge cases and error conditions gracefully
|
||||
- Use context managers for resource management
|
||||
- Apply composition over inheritance
|
||||
- Consider security implications for all API endpoints
|
||||
- Optimize for performance while maintaining readability
|
||||
|
||||
**Testing & Validation:**
|
||||
- Write pytest tests that validate spec requirements
|
||||
- Include unit tests for business logic
|
||||
- Integration tests for API endpoints
|
||||
- Test error conditions and edge cases
|
||||
- Use fixtures for consistent test setup
|
||||
- Mock external dependencies appropriately
|
||||
|
||||
**Debugging & Problem Solving:**
|
||||
- Analyze error messages and stack traces methodically
|
||||
- Identify root causes rather than applying quick fixes
|
||||
- Use logging effectively for troubleshooting
|
||||
- Apply systematic debugging approaches
|
||||
- Document solutions for future reference
|
||||
|
||||
**Basic Memory Integration:**
|
||||
- Use `mcp__basic-memory__read_note` to read specifications
|
||||
- Use `mcp__basic-memory__edit_note` to update specs with progress
|
||||
- Document implementation patterns and decisions
|
||||
- Link related services and database schemas
|
||||
- Maintain implementation history and troubleshooting guides
|
||||
|
||||
**Communication Style:**
|
||||
- Focus on concrete implementation results and working code
|
||||
- Document technical decisions and trade-offs clearly
|
||||
- Ask specific questions about requirements and constraints
|
||||
- Provide clear status updates on implementation progress
|
||||
- Explain code choices and architectural patterns
|
||||
|
||||
**Deliverables:**
|
||||
- Working Python services that meet spec requirements
|
||||
- Updated specifications with implementation status
|
||||
- Comprehensive tests validating functionality
|
||||
- Clean, maintainable, type-safe Python code
|
||||
- Proper error handling and logging
|
||||
- Database migrations and schema updates
|
||||
|
||||
**Key Principles:**
|
||||
- Implement specifications faithfully and completely
|
||||
- Write clean, efficient, and maintainable Python code
|
||||
- Follow established patterns and conventions
|
||||
- Apply proper error handling and security practices
|
||||
- Test thoroughly and document implementation decisions
|
||||
- Balance performance with code clarity and maintainability
|
||||
|
||||
When handed a specification via `/spec implement`, you will read the spec, understand the requirements, implement the Python solution using appropriate patterns and frameworks, create tests to validate functionality, and update the spec with completion status and any implementation notes.
|
||||
@@ -0,0 +1,126 @@
|
||||
---
|
||||
name: system-architect
|
||||
description: System architect who designs and implements architectural solutions, creates ADRs, and applies software engineering principles to solve complex system design problems.
|
||||
model: sonnet
|
||||
color: blue
|
||||
---
|
||||
|
||||
You are a Senior System Architect who designs and implements architectural solutions for complex software systems. You have deep expertise in software engineering principles, system design, multi-tenant SaaS architecture, and the Basic Memory Cloud platform.
|
||||
|
||||
**Primary Role: Architectural Implementation Agent**
|
||||
You design system architecture and implement architectural decisions through code, configuration, and documentation. You read specs from basic-memory, create architectural solutions, and update specs with implementation progress.
|
||||
|
||||
**Core Responsibilities:**
|
||||
|
||||
**Specification Implementation:**
|
||||
- Read architectural specs using basic-memory MCP tools
|
||||
- Design and implement system architecture solutions
|
||||
- Create code scaffolding, service structure, and system interfaces
|
||||
- Update specs with architectural decisions and implementation status
|
||||
- Document ADRs (Architectural Decision Records) for significant choices
|
||||
|
||||
**Architectural Design & Implementation:**
|
||||
- Design multi-service system architectures
|
||||
- Implement service boundaries and communication patterns
|
||||
- Create database schemas and migration strategies
|
||||
- Design authentication and authorization systems
|
||||
- Implement infrastructure-as-code patterns
|
||||
|
||||
**System Implementation Process:**
|
||||
1. **Read Spec**: Use `mcp__basic-memory__read_note` to understand architectural requirements
|
||||
2. **Design Solution**: Apply architectural principles and patterns
|
||||
3. **Implement Structure**: Create service scaffolding, interfaces, configurations
|
||||
4. **Document Decisions**: Create ADRs documenting architectural choices
|
||||
5. **Update Spec**: Record implementation progress and decisions
|
||||
6. **Validate**: Ensure implementation meets spec success criteria
|
||||
|
||||
**Architectural Principles Applied:**
|
||||
- DRY (Don't Repeat Yourself) - Single sources of truth
|
||||
- KISS (Keep It Simple Stupid) - Favor simplicity over cleverness
|
||||
- YAGNI (You Aren't Gonna Need It) - Build only what's needed now
|
||||
- Principle of Least Astonishment - Intuitive system behavior
|
||||
- Separation of Concerns - Clear boundaries and responsibilities
|
||||
|
||||
**Basic Memory Cloud Expertise:**
|
||||
|
||||
**Multi-Service Architecture:**
|
||||
- **Cloud Service**: Tenant management, OAuth 2.1, DBOS workflows
|
||||
- **MCP Gateway**: JWT validation, tenant routing, MCP proxy
|
||||
- **Web App**: Vue.js frontend, OAuth flows, user interface
|
||||
- **API Service**: Per-tenant Basic Memory instances with MCP
|
||||
|
||||
**Multi-Tenant SaaS Patterns:**
|
||||
- **Tenant Isolation**: Infrastructure-level isolation with dedicated instances
|
||||
- **Database-per-tenant**: Isolated PostgreSQL databases
|
||||
- **Authentication**: JWT tokens with tenant-specific claims
|
||||
- **Provisioning**: DBOS workflows for durable operations
|
||||
- **Resource Management**: Fly.io machine lifecycle management
|
||||
|
||||
**Implementation Capabilities:**
|
||||
- FastAPI service structure and middleware
|
||||
- DBOS workflow implementation
|
||||
- Database schema design and migrations
|
||||
- JWT authentication and authorization
|
||||
- Fly.io deployment configuration
|
||||
- Service communication patterns
|
||||
|
||||
**Technical Implementation:**
|
||||
- Create service scaffolding and project structure
|
||||
- Implement authentication and authorization middleware
|
||||
- Design database schemas and relationships
|
||||
- Configure deployment and infrastructure
|
||||
- Implement monitoring and health checks
|
||||
- Create API interfaces and contracts
|
||||
|
||||
**Code Quality Standards:**
|
||||
- Follow established patterns and conventions
|
||||
- Implement proper error handling and logging
|
||||
- Design for scalability and maintainability
|
||||
- Apply security best practices
|
||||
- Create comprehensive tests for architectural components
|
||||
- Document system behavior and interfaces
|
||||
|
||||
**Decision Documentation:**
|
||||
- Create ADRs for significant architectural choices
|
||||
- Document trade-offs and alternative approaches considered
|
||||
- Maintain decision history and rationale
|
||||
- Link architectural decisions to implementation code
|
||||
- Update decisions when new information becomes available
|
||||
|
||||
**Basic Memory Integration:**
|
||||
- Use `mcp__basic-memory__read_note` to read architectural specs
|
||||
- Use `mcp__basic-memory__write_note` to create ADRs and architectural documentation
|
||||
- Use `mcp__basic-memory__edit_note` to update specs with implementation progress
|
||||
- Document architectural patterns and anti-patterns for reuse
|
||||
- Maintain searchable knowledge base of system design decisions
|
||||
|
||||
**Communication Style:**
|
||||
- Focus on implemented solutions and concrete architectural artifacts
|
||||
- Document decisions with clear rationale and trade-offs
|
||||
- Provide specific implementation guidance and code examples
|
||||
- Ask targeted questions about requirements and constraints
|
||||
- Explain architectural choices in terms of business and technical impact
|
||||
|
||||
**Deliverables:**
|
||||
- Working system architecture implementations
|
||||
- ADRs documenting architectural decisions
|
||||
- Service scaffolding and interface definitions
|
||||
- Database schemas and migration scripts
|
||||
- Configuration and deployment artifacts
|
||||
- Updated specifications with implementation status
|
||||
|
||||
**Anti-Patterns to Avoid:**
|
||||
- Premature optimization over correctness
|
||||
- Over-engineering for current needs
|
||||
- Building without clear requirements
|
||||
- Creating multiple sources of truth
|
||||
- Implementing solutions without understanding root causes
|
||||
|
||||
**Key Principles:**
|
||||
- Implement architectural decisions through working code
|
||||
- Document all significant decisions and trade-offs
|
||||
- Build systems that teams can understand and maintain
|
||||
- Apply proven patterns and avoid reinventing solutions
|
||||
- Balance current needs with long-term maintainability
|
||||
|
||||
When handed an architectural specification via `/spec implement`, you will read the spec, design the solution applying architectural principles, implement the necessary code and configuration, document decisions through ADRs, and update the spec with completion status and architectural notes.
|
||||
@@ -30,8 +30,7 @@ The justfile target handles:
|
||||
- ✅ Beta version format validation (supports b1, b2, rc1, etc.)
|
||||
- ✅ Git status and branch checks
|
||||
- ✅ Quality checks (`just check` - lint, format, type-check, tests)
|
||||
- ✅ Version update across all consolidated manifests via `just set-version` (Python
|
||||
package + Claude Code plugin/marketplaces + Codex plugin + Hermes + OpenClaw)
|
||||
- ✅ Version update in `src/basic_memory/__init__.py`
|
||||
- ✅ Automatic commit with proper message
|
||||
- ✅ Tag creation and pushing to GitHub
|
||||
- ✅ Beta release workflow trigger
|
||||
@@ -91,6 +90,6 @@ Monitor release: https://github.com/basicmachines-co/basic-memory/actions
|
||||
- Beta releases are pre-releases for testing new features
|
||||
- Automatically published to PyPI with pre-release flag
|
||||
- Uses the automated justfile target for consistency
|
||||
- Version is automatically updated across all consolidated manifests via `just set-version`
|
||||
- Version is automatically updated in `__init__.py`
|
||||
- Ideal for validating changes before stable release
|
||||
- Supports both beta (b1, b2) and release candidate (rc1, rc2) versions
|
||||
- Supports both beta (b1, b2) and release candidate (rc1, rc2) versions
|
||||
@@ -28,10 +28,7 @@ You are an expert release manager for the Basic Memory project. When the user ru
|
||||
|
||||
#### Documentation Validation
|
||||
1. **Changelog Check**
|
||||
- CHANGELOG.md contains entry for target version **already landed on `main`**
|
||||
(main only accepts changes via PR, so the changelog entry must go through
|
||||
its own PR before running the release; the recipe pre-flight-checks for a
|
||||
`## vX.Y.Z` heading)
|
||||
- CHANGELOG.md contains entry for target version
|
||||
- Entry includes all major features and fixes
|
||||
- Breaking changes are documented
|
||||
|
||||
@@ -44,15 +41,10 @@ just release <version>
|
||||
The justfile target handles:
|
||||
- ✅ Version format validation
|
||||
- ✅ Git status and branch checks
|
||||
- ✅ Changelog entry check (must already be on `main`)
|
||||
- ✅ Quality checks (`just lint` + `just typecheck`)
|
||||
- ✅ Version update across all consolidated manifests via `just set-version` (Python
|
||||
package + Claude Code plugin/marketplaces + Codex plugin + Hermes + OpenClaw)
|
||||
- ✅ Release PR: commits the bump on a `release/vX.Y.Z` branch, opens a PR
|
||||
(`chore(core): release vX.Y.Z`), and rebase-merges it — the `main` ruleset
|
||||
rejects direct pushes and the repo disallows merge commits
|
||||
- ✅ Tags the rebased bump commit on `main` (found by commit subject, since
|
||||
the rebase rewrites the SHA) and pushes the tag
|
||||
- ✅ Quality checks (`just check` - lint, format, type-check, tests)
|
||||
- ✅ Version update in `src/basic_memory/__init__.py`
|
||||
- ✅ Automatic commit with proper message
|
||||
- ✅ Tag creation and pushing to GitHub
|
||||
- ✅ Release workflow trigger (automatic on tag push)
|
||||
|
||||
The GitHub Actions workflow (`.github/workflows/release.yml`) then:
|
||||
@@ -86,80 +78,45 @@ The GitHub Actions workflow (`.github/workflows/release.yml`) then:
|
||||
2. Verify formula version matches release
|
||||
3. Test Homebrew installation: `brew install basicmachines-co/basic-memory/basic-memory`
|
||||
|
||||
#### MCP Registry Publication
|
||||
|
||||
After PyPI release is published, update the MCP registry:
|
||||
|
||||
1. **Verify PyPI Release**
|
||||
- Confirm package is live: https://pypi.org/project/basic-memory/<version>/
|
||||
- The `server.json` version was auto-updated by `just release`
|
||||
|
||||
2. **Publish to MCP Registry**
|
||||
```bash
|
||||
# from the basic-memory repo root
|
||||
mcp-publisher publish
|
||||
```
|
||||
|
||||
If not authenticated:
|
||||
```bash
|
||||
mcp-publisher login github
|
||||
# Follow device authentication flow
|
||||
mcp-publisher publish
|
||||
```
|
||||
|
||||
3. **Verify Publication**
|
||||
```bash
|
||||
curl "https://registry.modelcontextprotocol.io/v0.1/servers?search=basic-memory"
|
||||
```
|
||||
|
||||
**Note:** The `mcp-publisher` CLI can be installed via Homebrew (`brew install mcp-publisher`) or from GitHub releases.
|
||||
|
||||
#### Website Updates
|
||||
|
||||
**1. basicmemory.com** (sibling `basicmemory.com` repo —
|
||||
`basicmachines-co/basicmemory.com`, formerly `basicmachines.co`)
|
||||
- **No version bump needed.** The marketing site is an Astro + React app and
|
||||
carries **no hardcoded Basic Memory version number** anywhere in its UI
|
||||
(`hero.tsx` and the rest of the site have no version string). The old
|
||||
instruction to bump `src/components/sections/hero.tsx` is obsolete — that
|
||||
file no longer holds a version. Release announcements are dated blog posts,
|
||||
not an in-place edit.
|
||||
- **Skip entirely for patch releases.**
|
||||
- **Significant releases only — optional announcement post**:
|
||||
**1. basicmachines.co** (`/Users/drew/code/basicmachines.co`)
|
||||
- **Goal**: Update version number displayed on the homepage
|
||||
- **Location**: Search for "Basic Memory v0." in the codebase to find version displays
|
||||
- **What to update**:
|
||||
- Hero section heading that shows "Basic Memory v{VERSION}"
|
||||
- "What's New in v{VERSION}" section heading
|
||||
- Feature highlights array (look for array of features with title/description)
|
||||
- **Process**:
|
||||
1. Pull latest from GitHub: `git pull origin main`
|
||||
2. Create release branch: `git checkout -b release/v{VERSION}`
|
||||
3. Add a dated post under `src/content/blog/` modeled on an existing
|
||||
release post (e.g. `basic-memory-v0-19-0-release.md`), summarizing 3–5
|
||||
headline features from `CHANGELOG.md`
|
||||
4. Commit (`git commit -s -m "..."`), push, and open a PR against
|
||||
`basicmachines-co/basicmemory.com`
|
||||
- **Deploy**: follow that repo's deployment process.
|
||||
3. Search codebase for current version number (e.g., "v0.16.1")
|
||||
4. Update version numbers to new release version
|
||||
5. Update feature highlights with 3-5 key features from this release (extract from CHANGELOG.md)
|
||||
6. Commit changes: `git commit -m "chore: update to v{VERSION}"`
|
||||
7. Push branch: `git push origin release/v{VERSION}`
|
||||
- **Deploy**: Follow deployment process for basicmachines.co
|
||||
|
||||
**2. docs.basicmemory.com** (sibling `docs.basicmemory.com` repo)
|
||||
- **Goal**: Add a What's New page for the release and bump the homepage badge
|
||||
- **Site shape**: Nuxt/Docus content site. The changelog page
|
||||
(`content/2.whats-new/*.changelog.md`) auto-fetches GitHub releases — no
|
||||
manual changelog update needed. See that repo's CLAUDE.md "Version Bump
|
||||
Checklist".
|
||||
**2. docs.basicmemory.com** (`/Users/drew/code/docs.basicmemory.com`)
|
||||
- **Goal**: Add new release notes section to the latest-releases page
|
||||
- **File**: `src/pages/latest-releases.mdx`
|
||||
- **What to do**:
|
||||
1. Pull latest from GitHub: `git pull origin main`
|
||||
2. Create release branch: `git checkout -b release/v{VERSION}`
|
||||
3. Read `CHANGELOG.md` in the `basic-memory` repo to get release content
|
||||
4. **New minor/major release**: add `content/2.whats-new/1.v{VERSION}.md`
|
||||
modeled on the previous version page (frontmatter title/description,
|
||||
headline feature first, then sections, then an Upgrading note) and
|
||||
renumber the existing what's-new pages down one slot (URLs don't
|
||||
change — Nuxt strips the numeric prefixes)
|
||||
5. **Patch release**: append a short note to the current version's page
|
||||
instead of creating a new one
|
||||
6. Update the homepage version badge in `content/index.md` (the
|
||||
`v0.XX →` button text and its `to: /whats-new/v{VERSION}` link)
|
||||
7. If the release adds user-facing features, update the relevant guide
|
||||
and reference pages (`content/3.cloud/`, `content/9.reference/`)
|
||||
8. Commit: `git commit -s -m "docs: add v{VERSION} release notes"`
|
||||
9. Push branch and open a PR; merge after the release is tagged
|
||||
- **Deploy**: push to main auto-deploys to development; production requires
|
||||
manual workflow dispatch via GitHub Actions
|
||||
3. Read the existing file to understand the format and structure
|
||||
4. Read `/Users/drew/code/basic-memory/CHANGELOG.md` to get release content
|
||||
5. Add new release section **at the top** (after MDX imports, before other releases)
|
||||
6. Follow the existing pattern:
|
||||
- Heading: `## [v{VERSION}](github-link) — YYYY-MM-DD`
|
||||
- Focus statement if applicable
|
||||
- `<Info>` block with highlights (3-5 key items)
|
||||
- Sections for Features, Bug Fixes, Breaking Changes, etc.
|
||||
- Link to full changelog at the end
|
||||
- Separator `---` between releases
|
||||
7. Commit changes: `git commit -m "docs: add v{VERSION} release notes"`
|
||||
8. Push branch: `git push origin release/v{VERSION}`
|
||||
- **Source content**: Extract and format sections from CHANGELOG.md for this version
|
||||
- **Deploy**: Follow deployment process for docs.basicmemory.com
|
||||
|
||||
**4. Announce Release**
|
||||
- Post to Discord community if significant changes
|
||||
@@ -188,7 +145,6 @@ Before starting, verify:
|
||||
📋 GitHub Release: https://github.com/basicmachines-co/basic-memory/releases/tag/v0.13.2
|
||||
📦 PyPI: https://pypi.org/project/basic-memory/0.13.2/
|
||||
🍺 Homebrew: https://github.com/basicmachines-co/homebrew-basic-memory
|
||||
🔌 MCP Registry: https://registry.modelcontextprotocol.io
|
||||
🚀 GitHub Actions: Completed
|
||||
|
||||
Install with pip/uv:
|
||||
@@ -206,16 +162,8 @@ Users can now upgrade:
|
||||
- This creates production releases used by end users
|
||||
- Must pass all quality gates before proceeding
|
||||
- Uses the automated justfile target for consistency
|
||||
- Version is automatically updated across **all** consolidated manifests via
|
||||
`just set-version <version>` (which calls `scripts/update_versions.py`): the
|
||||
Python package (`__init__.py`, `server.json`) **and** the plugin/agent artifacts
|
||||
(Claude Code `plugin.json` + root/local marketplaces, Codex `plugin.json`,
|
||||
Hermes `plugin.yaml` + `__init__.py`, OpenClaw `package.json`). To bump only
|
||||
the plugin/agent artifacts
|
||||
out of band, use `just set-packages-version <version>` (preview with
|
||||
`just set-packages-version-dry-run <version>`).
|
||||
- Version is automatically updated in `__init__.py`
|
||||
- Triggers automated GitHub release with changelog
|
||||
- Package is published to PyPI for `pip` and `uv` users
|
||||
- Homebrew formula is automatically updated for stable releases
|
||||
- MCP Registry is updated manually via `mcp-publisher publish`
|
||||
- Supports multiple installation methods (uv, pip, Homebrew)
|
||||
- Supports multiple installation methods (uv, pip, Homebrew)
|
||||
@@ -1,21 +0,0 @@
|
||||
{
|
||||
"$schema": "https://json.schemastore.org/claude-code-settings.json",
|
||||
"env": {
|
||||
"CLAUDE_BASH_MAINTAIN_PROJECT_WORKING_DIR": "1",
|
||||
"CLAUDE_CODE_DISABLE_FEEDBACK_SURVEY": "1",
|
||||
"CLAUDE_CODE_NO_FLICKER": "1",
|
||||
"CLAUDE_CODE_DISABLE_ADAPTIVE_THINKING": "1"
|
||||
},
|
||||
"permissions": {
|
||||
"allow": [
|
||||
"Bash(just fast-check)",
|
||||
"Bash(just check)",
|
||||
"Bash(just fix)",
|
||||
"Bash(just typecheck)",
|
||||
"Bash(just lint)",
|
||||
"Bash(just test)"
|
||||
],
|
||||
"deny": []
|
||||
},
|
||||
"enableAllProjectMcpServers": true
|
||||
}
|
||||
@@ -1 +0,0 @@
|
||||
../../.agents/skills/adversarial-review
|
||||
@@ -1 +0,0 @@
|
||||
../../.agents/skills/basic-machines-review
|
||||
@@ -1 +0,0 @@
|
||||
../../.agents/skills/instrumentation
|
||||
@@ -1,25 +0,0 @@
|
||||
# Auto BM Soul
|
||||
|
||||
Write project updates for humans who will return later trying to understand what happened.
|
||||
|
||||
## Voice
|
||||
|
||||
- Clear, direct, warm, and technically honest.
|
||||
- Prefer concrete observations over generic praise.
|
||||
- It is okay to say when code is messy, risky, clever, boring, or satisfying.
|
||||
- Keep personality in service of memory, not performance.
|
||||
|
||||
## Do
|
||||
|
||||
- Tell the story.
|
||||
- Name the tradeoffs.
|
||||
- Call out sharp edges.
|
||||
- Notice good simplifications.
|
||||
- Let the note have taste and a little life when the evidence supports it.
|
||||
|
||||
## Do Not
|
||||
|
||||
- Do not invent intent, impact, tests, or drama.
|
||||
- Dunk on people.
|
||||
- Turn the note into marketing copy.
|
||||
- Hide uncertainty behind confident prose.
|
||||
@@ -1,7 +0,0 @@
|
||||
project: dev
|
||||
workspace: basic-memory-7020de4e925843c68c9056c60d101d9e
|
||||
deploy_workflows:
|
||||
- Deploy Production
|
||||
production_environments:
|
||||
- production
|
||||
note_folder: project-updates/github/{owner}/{repo}
|
||||
@@ -1,64 +0,0 @@
|
||||
# Memory CI Capture
|
||||
|
||||
You turn GitHub delivery context into a durable project update for Basic Memory.
|
||||
GitHub records the mechanics. Basic Memory remembers what changed and why.
|
||||
|
||||
## Inputs
|
||||
|
||||
- Read `.github/basic-memory/project-update-context.json`.
|
||||
- Read `.github/basic-memory/SOUL.md` if it exists. It is the repo-local voice and style guide
|
||||
for project updates.
|
||||
- Read the PR diff before writing when a SHA is available. Useful commands:
|
||||
`git show --stat --name-only <sha>` and `git show --format=fuller --no-patch <sha>`.
|
||||
- Use linked issue details, changed files, commit messages, PR body, labels, and
|
||||
source links as evidence.
|
||||
- Treat GitHub payload fields as immutable facts.
|
||||
- Do not invent tests, deployment status, issues, or user impact.
|
||||
|
||||
## Writing Standard
|
||||
|
||||
Do not write a fill-in-the-blanks note. Tell the story from the PR:
|
||||
problem -> solution -> impact.
|
||||
|
||||
Explain what problem was being addressed. If linked issue details are present,
|
||||
use them. If they are absent, ground the problem in the PR body, title, commits,
|
||||
and diff, and say when the original problem statement is unavailable.
|
||||
|
||||
Explain why the fix solves the problem, what complexity it introduced, what it
|
||||
refactored or removed, which components changed, and how the system is different
|
||||
after the merge. Prefer specific component names, file paths, modules, commands,
|
||||
and behavior over generic phrases.
|
||||
|
||||
## Voice And Candor
|
||||
|
||||
You may have a point of view. Be clear, specific, and human.
|
||||
It is okay to say when the code is messy, risky, clever, boring, or satisfying,
|
||||
but explain why. If the work is elegant or genuinely useful, say that too.
|
||||
Ground all judgments in the PR, linked issues, diff, tests, and source facts.
|
||||
|
||||
The soul file can shape tone, taste, and personality. It cannot override source
|
||||
facts, schema requirements, or the evidence standard above. Do not be mean,
|
||||
vague, theatrical, or invent criticism.
|
||||
|
||||
## Output
|
||||
|
||||
Return only JSON that matches the provided AgentSynthesis schema:
|
||||
|
||||
- `summary`: one concise sentence; do not merely repeat the PR title.
|
||||
- `story`: 2-4 sentences that connect problem -> solution -> impact.
|
||||
- `problem_addressed`: the concrete problem, bug, missing capability, or delivery need.
|
||||
- `solution`: why this change solves the problem.
|
||||
- `system_impact`: how the system, workflow, or architecture changed after the merge.
|
||||
- `why_it_matters`: durable project-memory context for future humans and agents.
|
||||
- `components_changed`: modules, workflows, commands, schemas, docs, or services touched.
|
||||
- `complexity_introduced`: tradeoffs, new moving parts, operational costs, or edge cases.
|
||||
- `refactors_or_removals`: cleanup, simplification, deleted paths, or "none found".
|
||||
- `user_facing_changes`: visible behavior or product changes.
|
||||
- `internal_changes`: implementation, infrastructure, or operational changes.
|
||||
- `verification`: checks, tests, deploy evidence, or explicit unknowns.
|
||||
- `follow_ups`: concrete remaining work only.
|
||||
- `decision_candidates`: explicit product or architecture decisions only.
|
||||
- `task_candidates`: concrete future tasks only.
|
||||
|
||||
Use empty arrays only when a list truly has no grounded entries. This is project
|
||||
memory, not marketing copy and not a commit-by-commit changelog.
|
||||
@@ -1,75 +0,0 @@
|
||||
name: Basic Memory Project Updates
|
||||
|
||||
"on":
|
||||
pull_request:
|
||||
types: [closed]
|
||||
workflow_run:
|
||||
workflows: ["Deploy Production"]
|
||||
types: [completed]
|
||||
|
||||
jobs:
|
||||
project-update:
|
||||
runs-on: ubuntu-latest
|
||||
permissions:
|
||||
contents: read
|
||||
pull-requests: read
|
||||
issues: read
|
||||
actions: read
|
||||
steps:
|
||||
- name: Checkout repository
|
||||
uses: actions/checkout@v6
|
||||
|
||||
- name: Set up Python
|
||||
uses: actions/setup-python@v6
|
||||
with:
|
||||
python-version: "3.12"
|
||||
|
||||
- name: Install Basic Memory from checkout
|
||||
run: |
|
||||
python -m pip install --upgrade pip
|
||||
pip install -e .
|
||||
|
||||
- name: Collect project update context
|
||||
id: collect
|
||||
env:
|
||||
GITHUB_TOKEN: ${{ github.token }}
|
||||
run: |
|
||||
bm ci collect \
|
||||
--config .github/basic-memory/config.yml \
|
||||
--output .github/basic-memory/project-update-context.json
|
||||
|
||||
- name: Stop when event is not eligible
|
||||
if: steps.collect.outputs.eligible != 'true'
|
||||
run: |
|
||||
echo "Auto BM skipped: ${{ steps.collect.outputs.skip_reason }}"
|
||||
|
||||
- name: Write Codex output schema
|
||||
if: steps.collect.outputs.eligible == 'true'
|
||||
run: |
|
||||
bm ci agent-schema --output "${{ runner.temp }}/agent-synthesis.schema.json"
|
||||
|
||||
- name: Synthesize project update with Codex
|
||||
if: steps.collect.outputs.eligible == 'true'
|
||||
uses: openai/codex-action@v1
|
||||
with:
|
||||
openai-api-key: ${{ secrets.OPENAI_API_KEY }}
|
||||
prompt-file: .github/basic-memory/memory-ci-capture.md
|
||||
output-file: ${{ runner.temp }}/agent-synthesis.json
|
||||
output-schema-file: ${{ runner.temp }}/agent-synthesis.schema.json
|
||||
sandbox: read-only
|
||||
safety-strategy: drop-sudo
|
||||
|
||||
- name: Publish project update
|
||||
if: steps.collect.outputs.eligible == 'true'
|
||||
env:
|
||||
BASIC_MEMORY_CLOUD_API_KEY: ${{ secrets.BASIC_MEMORY_API_KEY }}
|
||||
BASIC_MEMORY_CI_CLOUD_HOST: ${{ vars.BASIC_MEMORY_CLOUD_HOST }}
|
||||
run: |
|
||||
if [ -n "$BASIC_MEMORY_CI_CLOUD_HOST" ]; then
|
||||
export BASIC_MEMORY_CLOUD_HOST="$BASIC_MEMORY_CI_CLOUD_HOST"
|
||||
fi
|
||||
bm ci publish \
|
||||
--cloud \
|
||||
--config .github/basic-memory/config.yml \
|
||||
--context .github/basic-memory/project-update-context.json \
|
||||
--synthesis "${{ runner.temp }}/agent-synthesis.json"
|
||||
@@ -1,18 +1,23 @@
|
||||
name: Claude Code Review
|
||||
|
||||
"on":
|
||||
workflow_dispatch:
|
||||
inputs:
|
||||
pr_number:
|
||||
description: Pull request number to review manually
|
||||
required: true
|
||||
|
||||
env:
|
||||
FORCE_JAVASCRIPT_ACTIONS_TO_NODE24: "true"
|
||||
on:
|
||||
pull_request:
|
||||
types: [opened, synchronize]
|
||||
# Optional: Only run on specific file changes
|
||||
# paths:
|
||||
# - "src/**/*.ts"
|
||||
# - "src/**/*.tsx"
|
||||
# - "src/**/*.js"
|
||||
# - "src/**/*.jsx"
|
||||
|
||||
jobs:
|
||||
claude-review:
|
||||
if: inputs.pr_number != ''
|
||||
# Only run for organization members and collaborators
|
||||
if: |
|
||||
github.event.pull_request.author_association == 'OWNER' ||
|
||||
github.event.pull_request.author_association == 'MEMBER' ||
|
||||
github.event.pull_request.author_association == 'COLLABORATOR'
|
||||
|
||||
runs-on: ubuntu-latest
|
||||
permissions:
|
||||
contents: read
|
||||
@@ -22,7 +27,7 @@ jobs:
|
||||
|
||||
steps:
|
||||
- name: Checkout repository
|
||||
uses: actions/checkout@v6
|
||||
uses: actions/checkout@v4
|
||||
with:
|
||||
fetch-depth: 1
|
||||
|
||||
@@ -35,14 +40,7 @@ jobs:
|
||||
track_progress: true # Enable visual progress tracking
|
||||
allowed_bots: '*'
|
||||
prompt: |
|
||||
Review Basic Memory PR #${{ inputs.pr_number }} as an advisory manual review.
|
||||
|
||||
Use `gh pr view ${{ inputs.pr_number }}` and related `gh pr`/`gh api`
|
||||
commands to inspect the pull request. Do not merge the PR and do not
|
||||
treat this advisory review as the required merge gate. BM Bossbot owns
|
||||
the required `BM Bossbot Approval` status.
|
||||
|
||||
Review the PR against our team checklist:
|
||||
Review this Basic Memory PR against our team checklist:
|
||||
|
||||
## Code Quality & Standards
|
||||
- [ ] Follows Basic Memory's coding conventions in CLAUDE.md
|
||||
@@ -56,7 +54,6 @@ jobs:
|
||||
- [ ] Unit tests for new functions/methods
|
||||
- [ ] Integration tests for new MCP tools
|
||||
- [ ] Test coverage for edge cases
|
||||
- [ ] **100% test coverage maintained** (use `# pragma: no cover` only for truly hard-to-test code)
|
||||
- [ ] Documentation updated (README, docstrings)
|
||||
- [ ] CLAUDE.md updated if conventions change
|
||||
|
||||
|
||||
@@ -4,9 +4,6 @@ on:
|
||||
issues:
|
||||
types: [opened]
|
||||
|
||||
env:
|
||||
FORCE_JAVASCRIPT_ACTIONS_TO_NODE24: "true"
|
||||
|
||||
jobs:
|
||||
triage:
|
||||
runs-on: ubuntu-latest
|
||||
@@ -15,7 +12,7 @@ jobs:
|
||||
id-token: write
|
||||
steps:
|
||||
- name: Checkout repository
|
||||
uses: actions/checkout@v6
|
||||
uses: actions/checkout@v4
|
||||
with:
|
||||
fetch-depth: 1
|
||||
|
||||
@@ -71,4 +68,4 @@ jobs:
|
||||
|
||||
Read the issue carefully and provide helpful triage with appropriate labels.
|
||||
|
||||
claude_args: '--allowed-tools "Bash(gh issue:*),Bash(gh search:*),Read"'
|
||||
claude_args: '--allowed-tools "Bash(gh issue:*),Bash(gh search:*),Read"'
|
||||
@@ -12,9 +12,6 @@ on:
|
||||
pull_request_target:
|
||||
types: [opened, synchronize]
|
||||
|
||||
env:
|
||||
FORCE_JAVASCRIPT_ACTIONS_TO_NODE24: "true"
|
||||
|
||||
jobs:
|
||||
claude:
|
||||
if: |
|
||||
@@ -44,7 +41,7 @@ jobs:
|
||||
actions: read # Required for Claude to read CI results on PRs
|
||||
steps:
|
||||
- name: Checkout repository
|
||||
uses: actions/checkout@v6
|
||||
uses: actions/checkout@v4
|
||||
with:
|
||||
# For pull_request_target, checkout the PR head to review the actual changes
|
||||
ref: ${{ github.event_name == 'pull_request_target' && github.event.pull_request.head.sha || github.sha }}
|
||||
@@ -68,3 +65,4 @@ jobs:
|
||||
# See https://github.com/anthropics/claude-code-action/blob/main/docs/usage.md
|
||||
# or https://docs.claude.com/en/docs/claude-code/sdk#command-line for available options
|
||||
# claude_args: '--model claude-opus-4-1-20250805 --allowed-tools Bash(gh pr:*)'
|
||||
|
||||
|
||||
@@ -1,110 +0,0 @@
|
||||
name: Consolidated Packages
|
||||
|
||||
on:
|
||||
pull_request:
|
||||
paths:
|
||||
- ".claude-plugin/**"
|
||||
- "plugins/**"
|
||||
- "skills/**"
|
||||
- "integrations/**"
|
||||
- "scripts/update_versions.py"
|
||||
- "scripts/validate_*.py"
|
||||
- "justfile"
|
||||
- ".github/workflows/consolidated-packages.yml"
|
||||
push:
|
||||
branches:
|
||||
- main
|
||||
paths:
|
||||
- ".claude-plugin/**"
|
||||
- "plugins/**"
|
||||
- "skills/**"
|
||||
- "integrations/**"
|
||||
- "scripts/update_versions.py"
|
||||
- "scripts/validate_*.py"
|
||||
- "justfile"
|
||||
- ".github/workflows/consolidated-packages.yml"
|
||||
|
||||
permissions:
|
||||
contents: read
|
||||
|
||||
jobs:
|
||||
claude-code:
|
||||
name: Claude Code marketplace
|
||||
permissions:
|
||||
contents: read
|
||||
runs-on: ubuntu-latest
|
||||
steps:
|
||||
- uses: actions/checkout@v4
|
||||
|
||||
- uses: extractions/setup-just@v4
|
||||
|
||||
- name: Validate manifests, agent, and bundled skills
|
||||
run: just --justfile plugins/claude-code/justfile --working-directory plugins/claude-code ci-check
|
||||
|
||||
- name: Verify shared version dry run
|
||||
run: just release-dry-run v0.99.0
|
||||
|
||||
skills:
|
||||
name: Shared skills
|
||||
permissions:
|
||||
contents: read
|
||||
runs-on: ubuntu-latest
|
||||
steps:
|
||||
- uses: actions/checkout@v4
|
||||
|
||||
- uses: extractions/setup-just@v4
|
||||
|
||||
- name: Validate SKILL.md source
|
||||
run: just --justfile skills/justfile --working-directory skills check
|
||||
|
||||
hermes:
|
||||
name: Hermes unit tests
|
||||
permissions:
|
||||
contents: read
|
||||
runs-on: ubuntu-latest
|
||||
defaults:
|
||||
run:
|
||||
working-directory: integrations/hermes
|
||||
steps:
|
||||
- uses: actions/checkout@v4
|
||||
|
||||
- uses: extractions/setup-just@v4
|
||||
|
||||
- name: Install uv
|
||||
uses: astral-sh/setup-uv@v3
|
||||
|
||||
- name: Set up Python
|
||||
run: uv python install 3.12
|
||||
|
||||
- name: Validate manifest and run unit tests
|
||||
run: just check
|
||||
|
||||
openclaw:
|
||||
name: OpenClaw package
|
||||
permissions:
|
||||
contents: read
|
||||
runs-on: ubuntu-latest
|
||||
defaults:
|
||||
run:
|
||||
working-directory: integrations/openclaw
|
||||
steps:
|
||||
- uses: actions/checkout@v4
|
||||
|
||||
- uses: extractions/setup-just@v4
|
||||
|
||||
- name: Setup Bun
|
||||
uses: oven-sh/setup-bun@v2
|
||||
with:
|
||||
bun-version: "1.3.8"
|
||||
|
||||
- name: Setup Node
|
||||
uses: actions/setup-node@v4
|
||||
with:
|
||||
node-version: "24"
|
||||
registry-url: "https://registry.npmjs.org"
|
||||
|
||||
- name: Install dependencies
|
||||
run: bun install --frozen-lockfile
|
||||
|
||||
- name: Release readiness
|
||||
run: just release-check
|
||||
@@ -5,9 +5,6 @@ on:
|
||||
branches: [main]
|
||||
workflow_dispatch: # Allow manual triggering
|
||||
|
||||
env:
|
||||
FORCE_JAVASCRIPT_ACTIONS_TO_NODE24: "true"
|
||||
|
||||
jobs:
|
||||
dev-release:
|
||||
runs-on: ubuntu-latest
|
||||
@@ -16,12 +13,12 @@ jobs:
|
||||
contents: write
|
||||
|
||||
steps:
|
||||
- uses: actions/checkout@v6
|
||||
- uses: actions/checkout@v4
|
||||
with:
|
||||
fetch-depth: 0
|
||||
|
||||
- name: Set up Python
|
||||
uses: actions/setup-python@v6
|
||||
uses: actions/setup-python@v5
|
||||
with:
|
||||
python-version: "3.12"
|
||||
|
||||
@@ -53,4 +50,4 @@ jobs:
|
||||
uses: pypa/gh-action-pypi-publish@release/v1
|
||||
with:
|
||||
password: ${{ secrets.PYPI_TOKEN }}
|
||||
skip-existing: true # Don't fail if version already exists
|
||||
skip-existing: true # Don't fail if version already exists
|
||||
@@ -9,27 +9,27 @@ on:
|
||||
env:
|
||||
REGISTRY: ghcr.io
|
||||
IMAGE_NAME: basicmachines-co/basic-memory
|
||||
FORCE_JAVASCRIPT_ACTIONS_TO_NODE24: "true"
|
||||
|
||||
jobs:
|
||||
docker:
|
||||
runs-on: depot-ubuntu-24.04
|
||||
runs-on: ubuntu-latest
|
||||
permissions:
|
||||
contents: read
|
||||
id-token: write
|
||||
packages: write
|
||||
|
||||
steps:
|
||||
- name: Checkout repository
|
||||
uses: actions/checkout@v6
|
||||
uses: actions/checkout@v4
|
||||
with:
|
||||
fetch-depth: 0
|
||||
|
||||
- name: Set up Depot
|
||||
uses: depot/setup-action@v1
|
||||
- name: Set up Docker Buildx
|
||||
uses: docker/setup-buildx-action@v3
|
||||
with:
|
||||
platforms: linux/amd64,linux/arm64
|
||||
|
||||
- name: Log in to GitHub Container Registry
|
||||
uses: docker/login-action@v4
|
||||
uses: docker/login-action@v3
|
||||
with:
|
||||
registry: ${{ env.REGISTRY }}
|
||||
username: ${{ github.actor }}
|
||||
@@ -37,7 +37,7 @@ jobs:
|
||||
|
||||
- name: Extract metadata
|
||||
id: meta
|
||||
uses: docker/metadata-action@v6
|
||||
uses: docker/metadata-action@v5
|
||||
with:
|
||||
images: ${{ env.REGISTRY }}/${{ env.IMAGE_NAME }}
|
||||
tags: |
|
||||
@@ -48,12 +48,14 @@ jobs:
|
||||
type=raw,value=latest,enable={{is_default_branch}}
|
||||
|
||||
- name: Build and push Docker image
|
||||
uses: depot/build-push-action@v1
|
||||
uses: docker/build-push-action@v5
|
||||
with:
|
||||
project: ${{ vars.DEPOT_BASIC_MEMORY_PROJECT_ID || vars.DEPOT_PROJECT_ID }}
|
||||
context: .
|
||||
file: ./Dockerfile
|
||||
platforms: linux/amd64,linux/arm64
|
||||
push: true
|
||||
tags: ${{ steps.meta.outputs.tags }}
|
||||
labels: ${{ steps.meta.outputs.labels }}
|
||||
cache-from: type=gha
|
||||
cache-to: type=gha,mode=max
|
||||
|
||||
|
||||
@@ -7,14 +7,11 @@ on:
|
||||
- edited
|
||||
- synchronize
|
||||
|
||||
env:
|
||||
FORCE_JAVASCRIPT_ACTIONS_TO_NODE24: "true"
|
||||
|
||||
jobs:
|
||||
main:
|
||||
runs-on: ubuntu-latest
|
||||
steps:
|
||||
- uses: amannn/action-semantic-pull-request@v6
|
||||
- uses: amannn/action-semantic-pull-request@v5
|
||||
env:
|
||||
GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
|
||||
with:
|
||||
@@ -38,11 +35,7 @@ jobs:
|
||||
mcp
|
||||
sync
|
||||
ui
|
||||
ci
|
||||
deps
|
||||
installer
|
||||
plugins
|
||||
skills
|
||||
integrations
|
||||
# Allow breaking changes (needs "!" after type/scope)
|
||||
requireScopeForBreakingChange: true
|
||||
requireScopeForBreakingChange: true
|
||||
+22
-102
@@ -5,9 +5,6 @@ on:
|
||||
tags:
|
||||
- 'v*' # Trigger on version tags like v1.0.0, v0.13.0, etc.
|
||||
|
||||
env:
|
||||
FORCE_JAVASCRIPT_ACTIONS_TO_NODE24: "true"
|
||||
|
||||
jobs:
|
||||
release:
|
||||
runs-on: ubuntu-latest
|
||||
@@ -16,12 +13,12 @@ jobs:
|
||||
contents: write
|
||||
|
||||
steps:
|
||||
- uses: actions/checkout@v6
|
||||
- uses: actions/checkout@v4
|
||||
with:
|
||||
fetch-depth: 0
|
||||
|
||||
- name: Set up Python
|
||||
uses: actions/setup-python@v6
|
||||
uses: actions/setup-python@v5
|
||||
with:
|
||||
python-version: "3.12"
|
||||
|
||||
@@ -42,7 +39,7 @@ jobs:
|
||||
echo "Build completed successfully"
|
||||
|
||||
- name: Create GitHub Release
|
||||
uses: softprops/action-gh-release@v3
|
||||
uses: softprops/action-gh-release@v2
|
||||
with:
|
||||
files: |
|
||||
dist/*.whl
|
||||
@@ -56,46 +53,6 @@ jobs:
|
||||
with:
|
||||
password: ${{ secrets.PYPI_TOKEN }}
|
||||
|
||||
openclaw:
|
||||
name: Publish OpenClaw npm Package
|
||||
needs: release
|
||||
# npm publishes only for stable product tags. Pre-release Python tags use
|
||||
# versions like 0.21.3b1, which are not valid npm pre-release semver.
|
||||
if: ${{ !contains(github.ref_name, 'dev') && !contains(github.ref_name, 'b') && !contains(github.ref_name, 'rc') }}
|
||||
runs-on: ubuntu-latest
|
||||
permissions:
|
||||
contents: read
|
||||
id-token: write
|
||||
defaults:
|
||||
run:
|
||||
working-directory: integrations/openclaw
|
||||
steps:
|
||||
- uses: actions/checkout@v6
|
||||
|
||||
- uses: extractions/setup-just@v4
|
||||
|
||||
- name: Setup Bun
|
||||
uses: oven-sh/setup-bun@v2
|
||||
with:
|
||||
bun-version: "1.3.8"
|
||||
|
||||
- name: Setup Node
|
||||
uses: actions/setup-node@v4
|
||||
with:
|
||||
node-version: "24"
|
||||
registry-url: "https://registry.npmjs.org"
|
||||
|
||||
- name: Install dependencies
|
||||
run: bun install --frozen-lockfile
|
||||
|
||||
- name: Release readiness
|
||||
run: just release-check
|
||||
|
||||
- name: Publish to npm
|
||||
run: npm publish --access public --provenance
|
||||
env:
|
||||
NODE_AUTH_TOKEN: ${{ secrets.NPM_TOKEN }}
|
||||
|
||||
homebrew:
|
||||
name: Update Homebrew Formula
|
||||
needs: release
|
||||
@@ -103,63 +60,26 @@ jobs:
|
||||
# Only run for stable releases (not dev, beta, or rc versions)
|
||||
if: ${{ !contains(github.ref_name, 'dev') && !contains(github.ref_name, 'b') && !contains(github.ref_name, 'rc') }}
|
||||
permissions:
|
||||
contents: read
|
||||
contents: write
|
||||
actions: read
|
||||
steps:
|
||||
# Inline bump replaces mislav/bump-homebrew-formula-action@v4.x.
|
||||
# The action does a HEAD request to api.github.com /repos/.../tarball/<ref>
|
||||
# with the bearer token and expects a 302 redirect. GitHub now returns
|
||||
# 303 on that endpoint when authenticated, which the action treats as a
|
||||
# fatal error. Re-implementing the bump as plain git+sed keeps the same
|
||||
# contract (update url + sha256, commit, push) with no third-party action.
|
||||
- name: Update Homebrew formula
|
||||
uses: mislav/bump-homebrew-formula-action@v3
|
||||
with:
|
||||
# Formula name in homebrew-basic-memory repo
|
||||
formula-name: basic-memory
|
||||
# The tap repository
|
||||
homebrew-tap: basicmachines-co/homebrew-basic-memory
|
||||
# Base branch of the tap repository
|
||||
base-branch: main
|
||||
# Download URL will be automatically constructed from the tag
|
||||
download-url: https://github.com/basicmachines-co/basic-memory/archive/refs/tags/${{ github.ref_name }}.tar.gz
|
||||
# Commit message for the formula update
|
||||
commit-message: |
|
||||
{{formulaName}} {{version}}
|
||||
|
||||
Created by https://github.com/basicmachines-co/basic-memory/actions/runs/${{ github.run_id }}
|
||||
env:
|
||||
HOMEBREW_TOKEN: ${{ secrets.HOMEBREW_TOKEN }}
|
||||
REF: ${{ github.ref_name }}
|
||||
REPO: ${{ github.repository }}
|
||||
RUN_URL: https://github.com/${{ github.repository }}/actions/runs/${{ github.run_id }}
|
||||
run: |
|
||||
set -euo pipefail
|
||||
# Personal Access Token with repo scope for homebrew-basic-memory repo
|
||||
COMMITTER_TOKEN: ${{ secrets.HOMEBREW_TOKEN }}
|
||||
|
||||
VERSION="${REF#v}"
|
||||
ARCHIVE_URL="https://github.com/${REPO}/archive/refs/tags/${REF}.tar.gz"
|
||||
|
||||
echo "::group::Compute tarball sha256"
|
||||
SHA256="$(curl --fail --silent --location "$ARCHIVE_URL" | sha256sum | awk '{print $1}')"
|
||||
test -n "$SHA256"
|
||||
echo "sha256: $SHA256"
|
||||
echo "::endgroup::"
|
||||
|
||||
echo "::group::Clone tap"
|
||||
git clone \
|
||||
--depth 1 \
|
||||
"https://x-access-token:${HOMEBREW_TOKEN}@github.com/basicmachines-co/homebrew-basic-memory.git" \
|
||||
tap
|
||||
cd tap
|
||||
git config user.name "github-actions[bot]"
|
||||
git config user.email "41898282+github-actions[bot]@users.noreply.github.com"
|
||||
echo "::endgroup::"
|
||||
|
||||
echo "::group::Patch Formula/basic-memory.rb"
|
||||
# Pipe-delimited sed because the URL contains slashes. The Formula
|
||||
# only has one `url` and one `sha256` directive, so a first-match
|
||||
# replacement is unambiguous. POSIX character classes ([[:space:]])
|
||||
# keep this portable across BSD and GNU sed.
|
||||
sed -i -E \
|
||||
-e "s|^([[:space:]]*url[[:space:]]+)\"[^\"]+\"|\1\"${ARCHIVE_URL}\"|" \
|
||||
-e "s|^([[:space:]]*sha256[[:space:]]+)\"[^\"]+\"|\1\"${SHA256}\"|" \
|
||||
Formula/basic-memory.rb
|
||||
git --no-pager diff Formula/basic-memory.rb
|
||||
echo "::endgroup::"
|
||||
|
||||
if git diff --quiet Formula/basic-memory.rb; then
|
||||
echo "Formula already at ${REF}; nothing to do."
|
||||
exit 0
|
||||
fi
|
||||
|
||||
echo "::group::Commit & push"
|
||||
git add Formula/basic-memory.rb
|
||||
git commit -m "basic-memory ${VERSION}
|
||||
|
||||
Created by ${RUN_URL}"
|
||||
git push origin HEAD:main
|
||||
echo "::endgroup::"
|
||||
|
||||
+57
-337
@@ -1,75 +1,53 @@
|
||||
name: Tests
|
||||
|
||||
concurrency:
|
||||
group: bm-ci-${{ github.workflow }}-${{ github.repository }}-${{ github.head_ref || github.ref }}
|
||||
cancel-in-progress: true
|
||||
|
||||
on:
|
||||
# Trigger: PR branch pushes already publish commit statuses that show up on the PR.
|
||||
# Why: running the full matrix on both push and pull_request doubles CI time for the
|
||||
# exact same branch head commit.
|
||||
# Outcome: each branch push runs the test suite once, including PR updates.
|
||||
push:
|
||||
|
||||
env:
|
||||
FORCE_JAVASCRIPT_ACTIONS_TO_NODE24: "true"
|
||||
# Branch builds (PRs arrive as push events — this workflow has no
|
||||
# pull_request trigger) select only impacted tests from the cached testmon
|
||||
# baseline (branch cache falling back to main's full-run recording). Pushes
|
||||
# to main run the full suite with --testmon-noselect to refresh the baseline.
|
||||
BASIC_MEMORY_TESTMON_FLAGS: ${{ github.ref_name == 'main' && '--testmon-noselect' || '--testmon --testmon-forceselect' }}
|
||||
branches: [ "main" ]
|
||||
pull_request:
|
||||
branches: [ "main" ]
|
||||
# pull_request_target runs on the BASE of the PR, not the merge result.
|
||||
# It has write permissions and access to secrets.
|
||||
# It's useful for PRs from forks or automated PRs but requires careful use for security reasons.
|
||||
# See: https://docs.github.com/en/actions/using-workflows/events-that-trigger-workflows#pull_request_target
|
||||
pull_request_target:
|
||||
branches: [ "main" ]
|
||||
|
||||
jobs:
|
||||
changes:
|
||||
# Docs/workflow-only changes skip the entire test matrix while the workflow
|
||||
# still concludes successfully, so the BM Bossbot gate (workflow_run on
|
||||
# Tests success) keeps firing and the PR stays mergeable.
|
||||
name: Detect code changes
|
||||
runs-on: ubuntu-latest
|
||||
outputs:
|
||||
code: ${{ steps.filter.outputs.code }}
|
||||
steps:
|
||||
- uses: actions/checkout@v6
|
||||
- id: filter
|
||||
uses: dorny/paths-filter@v3
|
||||
with:
|
||||
# Tests only runs on push events; for branch pushes compare against
|
||||
# main (merge-base), for main pushes dorny diffs the push range.
|
||||
base: main
|
||||
filters: |
|
||||
code:
|
||||
- 'src/**'
|
||||
- 'tests/**'
|
||||
- 'test-int/**'
|
||||
- 'alembic/**'
|
||||
- 'pyproject.toml'
|
||||
- 'uv.lock'
|
||||
- 'justfile'
|
||||
- '.github/workflows/test.yml'
|
||||
|
||||
static-checks:
|
||||
needs: changes
|
||||
if: needs.changes.outputs.code == 'true'
|
||||
name: Static Checks (Python 3.12)
|
||||
timeout-minutes: 20
|
||||
runs-on: ubuntu-latest
|
||||
test-sqlite:
|
||||
name: Test SQLite (${{ matrix.os }}, Python ${{ matrix.python-version }})
|
||||
strategy:
|
||||
fail-fast: false
|
||||
matrix:
|
||||
os: [ubuntu-latest, windows-latest]
|
||||
python-version: [ "3.12", "3.13" ]
|
||||
runs-on: ${{ matrix.os }}
|
||||
|
||||
steps:
|
||||
- uses: actions/checkout@v6
|
||||
- uses: actions/checkout@v4
|
||||
with:
|
||||
submodules: true
|
||||
|
||||
- name: Set up Python 3.12
|
||||
uses: actions/setup-python@v6
|
||||
- name: Set up Python ${{ matrix.python-version }}
|
||||
uses: actions/setup-python@v4
|
||||
with:
|
||||
python-version: "3.12"
|
||||
cache: "pip"
|
||||
python-version: ${{ matrix.python-version }}
|
||||
cache: 'pip'
|
||||
|
||||
- name: Install uv
|
||||
run: |
|
||||
pip install uv
|
||||
|
||||
- uses: extractions/setup-just@v4
|
||||
- name: Install just (Linux/macOS)
|
||||
if: runner.os != 'Windows'
|
||||
run: |
|
||||
curl --proto '=https' --tlsv1.2 -sSf https://just.systems/install.sh | bash -s -- --to /usr/local/bin
|
||||
|
||||
- name: Install just (Windows)
|
||||
if: runner.os == 'Windows'
|
||||
run: |
|
||||
# Install just using Chocolatey (pre-installed on GitHub Actions Windows runners)
|
||||
choco install just --yes
|
||||
shell: pwsh
|
||||
|
||||
- name: Create virtual env
|
||||
run: |
|
||||
@@ -77,7 +55,7 @@ jobs:
|
||||
|
||||
- name: Install dependencies
|
||||
run: |
|
||||
uv pip install -e ".[dev]"
|
||||
uv pip install -e .[dev]
|
||||
|
||||
- name: Run type checks
|
||||
run: |
|
||||
@@ -87,241 +65,42 @@ jobs:
|
||||
run: |
|
||||
just lint
|
||||
|
||||
test-sqlite-unit:
|
||||
needs: changes
|
||||
if: needs.changes.outputs.code == 'true'
|
||||
name: Test SQLite Unit (${{ matrix.os }}, Python ${{ matrix.python-version }})
|
||||
timeout-minutes: 45
|
||||
- name: Run tests (SQLite)
|
||||
run: |
|
||||
uv pip install pytest pytest-cov
|
||||
just test-sqlite
|
||||
|
||||
test-postgres:
|
||||
name: Test Postgres (Python ${{ matrix.python-version }})
|
||||
strategy:
|
||||
fail-fast: false
|
||||
matrix:
|
||||
include:
|
||||
- os: ubuntu-latest
|
||||
python-version: "3.12"
|
||||
- os: ubuntu-latest
|
||||
python-version: "3.13"
|
||||
# Python 3.14 unit tests are the longest full-suite slice; keep this
|
||||
# one on GitHub-hosted runners after Depot terminated it mid-suite.
|
||||
- os: ubuntu-latest
|
||||
python-version: "3.14"
|
||||
- os: windows-latest
|
||||
python-version: "3.12"
|
||||
runs-on: ${{ matrix.os }}
|
||||
|
||||
steps:
|
||||
- uses: actions/checkout@v6
|
||||
with:
|
||||
submodules: true
|
||||
|
||||
- name: Set up Python ${{ matrix.python-version }}
|
||||
uses: actions/setup-python@v6
|
||||
with:
|
||||
python-version: ${{ matrix.python-version }}
|
||||
cache: 'pip'
|
||||
|
||||
- name: Install uv
|
||||
run: |
|
||||
pip install uv
|
||||
|
||||
- uses: extractions/setup-just@v4
|
||||
|
||||
- name: Cache pytest-testmon results
|
||||
uses: actions/cache@v4
|
||||
with:
|
||||
path: |
|
||||
.testmondata
|
||||
.testmondata-shm
|
||||
.testmondata-wal
|
||||
key: ${{ runner.os }}-testmon-sqlite-unit-py${{ matrix.python-version }}-${{ github.ref_name }}-${{ github.run_id }}
|
||||
restore-keys: |
|
||||
${{ runner.os }}-testmon-sqlite-unit-py${{ matrix.python-version }}-${{ github.ref_name }}-
|
||||
${{ runner.os }}-testmon-sqlite-unit-py${{ matrix.python-version }}-main-
|
||||
${{ runner.os }}-testmon-sqlite-unit-py${{ matrix.python-version }}-
|
||||
|
||||
- name: Create virtual env
|
||||
run: |
|
||||
uv venv
|
||||
|
||||
- name: Install dependencies
|
||||
run: |
|
||||
uv pip install -e ".[dev]"
|
||||
|
||||
- name: Run tests
|
||||
run: |
|
||||
just test-unit-sqlite
|
||||
|
||||
test-sqlite-integration:
|
||||
needs: changes
|
||||
if: needs.changes.outputs.code == 'true'
|
||||
name: Test SQLite Integration (${{ matrix.os }}, Python ${{ matrix.python-version }})
|
||||
timeout-minutes: 45
|
||||
strategy:
|
||||
fail-fast: false
|
||||
matrix:
|
||||
include:
|
||||
- os: ubuntu-latest
|
||||
python-version: "3.12"
|
||||
- os: ubuntu-latest
|
||||
python-version: "3.13"
|
||||
- os: ubuntu-latest
|
||||
python-version: "3.14"
|
||||
- os: windows-latest
|
||||
python-version: "3.12"
|
||||
runs-on: ${{ matrix.os }}
|
||||
|
||||
steps:
|
||||
- uses: actions/checkout@v6
|
||||
with:
|
||||
submodules: true
|
||||
|
||||
- name: Set up Python ${{ matrix.python-version }}
|
||||
uses: actions/setup-python@v6
|
||||
with:
|
||||
python-version: ${{ matrix.python-version }}
|
||||
cache: 'pip'
|
||||
|
||||
- name: Install uv
|
||||
run: |
|
||||
pip install uv
|
||||
|
||||
- uses: extractions/setup-just@v4
|
||||
|
||||
- name: Cache pytest-testmon results
|
||||
uses: actions/cache@v4
|
||||
with:
|
||||
path: |
|
||||
.testmondata
|
||||
.testmondata-shm
|
||||
.testmondata-wal
|
||||
key: ${{ runner.os }}-testmon-sqlite-integration-py${{ matrix.python-version }}-${{ github.ref_name }}-${{ github.run_id }}
|
||||
restore-keys: |
|
||||
${{ runner.os }}-testmon-sqlite-integration-py${{ matrix.python-version }}-${{ github.ref_name }}-
|
||||
${{ runner.os }}-testmon-sqlite-integration-py${{ matrix.python-version }}-main-
|
||||
${{ runner.os }}-testmon-sqlite-integration-py${{ matrix.python-version }}-
|
||||
|
||||
- name: Create virtual env
|
||||
run: |
|
||||
uv venv
|
||||
|
||||
- name: Install dependencies
|
||||
run: |
|
||||
uv pip install -e ".[dev]"
|
||||
|
||||
- name: Run tests
|
||||
run: |
|
||||
just test-int-sqlite
|
||||
|
||||
test-postgres-unit:
|
||||
needs: changes
|
||||
if: needs.changes.outputs.code == 'true'
|
||||
name: Test Postgres Unit (Python ${{ matrix.python-version }}, shard ${{ matrix.group }}/3)
|
||||
timeout-minutes: 60
|
||||
strategy:
|
||||
fail-fast: false
|
||||
matrix:
|
||||
# Shard the largest suite across parallel jobs: each shard is a full job
|
||||
# with its own Postgres service running 1/3 of the collection.
|
||||
# Postgres runs on the latest Python only — the SQLite matrix carries
|
||||
# Python-version coverage; Postgres carries backend coverage.
|
||||
group: [1, 2, 3]
|
||||
python-version: ["3.14"]
|
||||
python-version: [ "3.12", "3.13" ]
|
||||
runs-on: ubuntu-latest
|
||||
|
||||
# Postgres service (only available on Linux runners)
|
||||
services:
|
||||
postgres:
|
||||
image: pgvector/pgvector:pg16
|
||||
image: postgres:17
|
||||
env:
|
||||
POSTGRES_DB: basic_memory_test
|
||||
POSTGRES_USER: basic_memory_user
|
||||
POSTGRES_PASSWORD: dev_password
|
||||
POSTGRES_DB: basic_memory_test
|
||||
ports:
|
||||
- 5432:5432
|
||||
options: >-
|
||||
--health-cmd "pg_isready -U basic_memory_user -d basic_memory_test"
|
||||
--health-cmd pg_isready
|
||||
--health-interval 10s
|
||||
--health-timeout 5s
|
||||
--health-retries 5
|
||||
env:
|
||||
BASIC_MEMORY_TEST_POSTGRES_URL: postgresql://basic_memory_user:dev_password@127.0.0.1:5432/basic_memory_test
|
||||
|
||||
steps:
|
||||
- uses: actions/checkout@v6
|
||||
with:
|
||||
submodules: true
|
||||
|
||||
- name: Set up Python ${{ matrix.python-version }}
|
||||
uses: actions/setup-python@v6
|
||||
with:
|
||||
python-version: ${{ matrix.python-version }}
|
||||
cache: 'pip'
|
||||
|
||||
- name: Install uv
|
||||
run: |
|
||||
pip install uv
|
||||
|
||||
- uses: extractions/setup-just@v4
|
||||
|
||||
- name: Cache pytest-testmon results
|
||||
uses: actions/cache@v4
|
||||
with:
|
||||
path: |
|
||||
.testmondata
|
||||
.testmondata-shm
|
||||
.testmondata-wal
|
||||
key: ${{ runner.os }}-testmon-postgres-unit-py${{ matrix.python-version }}-g${{ matrix.group }}-${{ github.ref_name }}-${{ github.run_id }}
|
||||
restore-keys: |
|
||||
${{ runner.os }}-testmon-postgres-unit-py${{ matrix.python-version }}-g${{ matrix.group }}-${{ github.ref_name }}-
|
||||
${{ runner.os }}-testmon-postgres-unit-py${{ matrix.python-version }}-g${{ matrix.group }}-main-
|
||||
${{ runner.os }}-testmon-postgres-unit-py${{ matrix.python-version }}-main-
|
||||
${{ runner.os }}-testmon-postgres-unit-py${{ matrix.python-version }}-
|
||||
|
||||
- name: Create virtual env
|
||||
run: |
|
||||
uv venv
|
||||
|
||||
- name: Install dependencies
|
||||
run: |
|
||||
uv pip install -e ".[dev]"
|
||||
|
||||
- name: Run tests
|
||||
run: |
|
||||
BASIC_MEMORY_PYTEST_SPLIT_FLAGS="--splits 3 --group ${{ matrix.group }}" just test-unit-postgres
|
||||
|
||||
test-postgres-integration:
|
||||
needs: changes
|
||||
if: needs.changes.outputs.code == 'true'
|
||||
name: Test Postgres Integration (Python ${{ matrix.python-version }})
|
||||
timeout-minutes: 45
|
||||
strategy:
|
||||
fail-fast: false
|
||||
matrix:
|
||||
# Latest Python only: SQLite carries version coverage, Postgres carries
|
||||
# backend coverage.
|
||||
python-version: ["3.14"]
|
||||
runs-on: ubuntu-latest
|
||||
services:
|
||||
postgres:
|
||||
image: pgvector/pgvector:pg16
|
||||
env:
|
||||
POSTGRES_USER: basic_memory_user
|
||||
POSTGRES_PASSWORD: dev_password
|
||||
POSTGRES_DB: basic_memory_test
|
||||
ports:
|
||||
- 5432:5432
|
||||
options: >-
|
||||
--health-cmd "pg_isready -U basic_memory_user -d basic_memory_test"
|
||||
--health-interval 10s
|
||||
--health-timeout 5s
|
||||
--health-retries 5
|
||||
env:
|
||||
BASIC_MEMORY_TEST_POSTGRES_URL: postgresql://basic_memory_user:dev_password@127.0.0.1:5432/basic_memory_test
|
||||
- 5433:5432
|
||||
|
||||
steps:
|
||||
- uses: actions/checkout@v6
|
||||
- uses: actions/checkout@v4
|
||||
with:
|
||||
submodules: true
|
||||
|
||||
- name: Set up Python ${{ matrix.python-version }}
|
||||
uses: actions/setup-python@v6
|
||||
uses: actions/setup-python@v4
|
||||
with:
|
||||
python-version: ${{ matrix.python-version }}
|
||||
cache: 'pip'
|
||||
@@ -330,20 +109,9 @@ jobs:
|
||||
run: |
|
||||
pip install uv
|
||||
|
||||
- uses: extractions/setup-just@v4
|
||||
|
||||
- name: Cache pytest-testmon results
|
||||
uses: actions/cache@v4
|
||||
with:
|
||||
path: |
|
||||
.testmondata
|
||||
.testmondata-shm
|
||||
.testmondata-wal
|
||||
key: ${{ runner.os }}-testmon-postgres-integration-py${{ matrix.python-version }}-${{ github.ref_name }}-${{ github.run_id }}
|
||||
restore-keys: |
|
||||
${{ runner.os }}-testmon-postgres-integration-py${{ matrix.python-version }}-${{ github.ref_name }}-
|
||||
${{ runner.os }}-testmon-postgres-integration-py${{ matrix.python-version }}-main-
|
||||
${{ runner.os }}-testmon-postgres-integration-py${{ matrix.python-version }}-
|
||||
- name: Install just
|
||||
run: |
|
||||
curl --proto '=https' --tlsv1.2 -sSf https://just.systems/install.sh | bash -s -- --to /usr/local/bin
|
||||
|
||||
- name: Create virtual env
|
||||
run: |
|
||||
@@ -351,57 +119,9 @@ jobs:
|
||||
|
||||
- name: Install dependencies
|
||||
run: |
|
||||
uv pip install -e ".[dev]"
|
||||
uv pip install -e .[dev]
|
||||
|
||||
- name: Run tests
|
||||
- name: Run tests (Postgres)
|
||||
run: |
|
||||
just test-int-postgres
|
||||
|
||||
test-semantic:
|
||||
needs: changes
|
||||
if: needs.changes.outputs.code == 'true'
|
||||
name: Test Semantic (Python 3.12)
|
||||
timeout-minutes: 45
|
||||
runs-on: ubuntu-latest
|
||||
|
||||
steps:
|
||||
- uses: actions/checkout@v6
|
||||
with:
|
||||
submodules: true
|
||||
|
||||
- name: Set up Python 3.12
|
||||
uses: actions/setup-python@v6
|
||||
with:
|
||||
python-version: "3.12"
|
||||
cache: "pip"
|
||||
|
||||
- name: Install uv
|
||||
run: |
|
||||
pip install uv
|
||||
|
||||
- uses: extractions/setup-just@v4
|
||||
|
||||
- name: Cache pytest-testmon results
|
||||
uses: actions/cache@v4
|
||||
with:
|
||||
path: |
|
||||
.testmondata
|
||||
.testmondata-shm
|
||||
.testmondata-wal
|
||||
key: ${{ runner.os }}-testmon-semantic-py3.12-${{ github.ref_name }}-${{ github.run_id }}
|
||||
restore-keys: |
|
||||
${{ runner.os }}-testmon-semantic-py3.12-${{ github.ref_name }}-
|
||||
${{ runner.os }}-testmon-semantic-py3.12-main-
|
||||
${{ runner.os }}-testmon-semantic-py3.12-
|
||||
|
||||
- name: Create virtual env
|
||||
run: |
|
||||
uv venv
|
||||
|
||||
- name: Install dependencies
|
||||
run: |
|
||||
uv pip install -e ".[dev]"
|
||||
|
||||
- name: Run tests
|
||||
run: |
|
||||
just test-semantic
|
||||
uv pip install pytest pytest-cov
|
||||
just test-postgres
|
||||
+2
-14
@@ -1,7 +1,6 @@
|
||||
*.py[cod]
|
||||
__pycache__/
|
||||
.pytest_cache/
|
||||
.testmondata*
|
||||
.coverage
|
||||
htmlcov/
|
||||
|
||||
@@ -49,19 +48,8 @@ ENV/
|
||||
/docs/.obsidian/
|
||||
/examples/.obsidian/
|
||||
/examples/.basic-memory/
|
||||
/docs/assets
|
||||
|
||||
|
||||
# claude action
|
||||
claude-output
|
||||
**/.claude/settings.local.json
|
||||
.mcp.json
|
||||
!/plugins/codex/.mcp.json
|
||||
.mcpregistry_*
|
||||
/.testmondata
|
||||
.benchmarks/
|
||||
|
||||
# Consolidated package build artifacts
|
||||
/integrations/openclaw/node_modules/
|
||||
/integrations/openclaw/dist/
|
||||
/integrations/openclaw/skills/
|
||||
/integrations/openclaw/*.tgz
|
||||
**/.claude/settings.local.json
|
||||
+1
-1
@@ -1 +1 @@
|
||||
3.14
|
||||
3.12
|
||||
|
||||
@@ -1,547 +0,0 @@
|
||||
# AGENTS.md - Basic Memory Project Guide
|
||||
|
||||
## Project Overview
|
||||
|
||||
Basic Memory is a local-first knowledge management system built on the Model Context Protocol (MCP). It enables
|
||||
bidirectional communication between LLMs (like Claude) and markdown files, creating a personal knowledge graph that can
|
||||
be traversed using links between documents.
|
||||
|
||||
## CODEBASE DEVELOPMENT
|
||||
|
||||
### Project information
|
||||
|
||||
See the [README.md](README.md) file for a project overview.
|
||||
|
||||
### Build and Test Commands
|
||||
|
||||
- Install: `just install` or `pip install -e ".[dev]"`
|
||||
- Run all tests (SQLite + Postgres): `just test`
|
||||
- Run all tests against SQLite: `just test-sqlite`
|
||||
- Run all tests against Postgres: `just test-postgres` (uses testcontainers)
|
||||
- Run unit tests (SQLite): `just test-unit-sqlite`
|
||||
- Run unit tests (Postgres): `just test-unit-postgres`
|
||||
- Run integration tests (SQLite): `just test-int-sqlite`
|
||||
- Run integration tests (Postgres): `just test-int-postgres`
|
||||
- Run impacted tests: `just testmon` (pytest-testmon; only tests affected by changed code)
|
||||
- Run MCP smoke test: `just test-smoke`
|
||||
- Fast local loop: `just fast-check` (default iteration flow)
|
||||
- Local consistency check: `just doctor`
|
||||
- Run all consolidated agent package checks: `just package-check`
|
||||
- Run Claude Code plugin checks: `just package-check-claude-code`
|
||||
- Run shared skills checks: `just package-check-skills`
|
||||
- Run Hermes plugin checks: `just package-check-hermes`
|
||||
- Run OpenClaw plugin checks: `just package-check-openclaw`
|
||||
- Run host-native agent harness checks: `just agent-harness-check`
|
||||
- Generate HTML coverage: `just coverage`
|
||||
- Single test: `pytest tests/path/to/test_file.py::test_function_name`
|
||||
- Run benchmarks: `pytest test-int/test_sync_performance_benchmark.py -v -m "benchmark and not slow"`
|
||||
- Lint: `just lint` or `ruff check . --fix`
|
||||
- Type check: `just typecheck` or `uv run ty check src tests test-int`
|
||||
- Type check (pyright): `just typecheck-pyright` or `uv run pyright`
|
||||
- Format: `just format` or `uv run ruff format .`
|
||||
- Run all code checks: `just check` (runs lint, format, typecheck, test)
|
||||
- Create db migration: `just migration "Your migration message"`
|
||||
- Run development MCP Inspector: `just run-inspector`
|
||||
|
||||
**Note:** Project requires Python 3.12+ (uses type parameter syntax and `type` aliases introduced in 3.12)
|
||||
|
||||
**Postgres Testing:** Uses [testcontainers](https://testcontainers-python.readthedocs.io/) which automatically spins up a Postgres instance in Docker. No manual database setup required - just have Docker running.
|
||||
|
||||
**Doctor Note:** `just doctor` runs with a temporary HOME/config so it won't touch your local Basic Memory settings. It leaves temp dirs in `/tmp` (safe to ignore or remove).
|
||||
|
||||
**Testmon Note:** When no files have changed, `just testmon` may collect 0 tests. That's expected and means no impacted tests were detected.
|
||||
|
||||
### Code/Test/Verify Loop (fast path)
|
||||
|
||||
1) **Code:** make changes.
|
||||
2) **Test:** `just fast-check` (lint/format/typecheck + pytest-testmon impacted tests for changed code).
|
||||
3) **Verify:** `just doctor` (end-to-end file ↔ DB loop in a temp project).
|
||||
4) **Package verify:** `just package-check` when changes touch `plugins/`, `skills/`, `integrations/`, package metadata, or release wiring.
|
||||
5) **Full gate (when needed):** `just test` or `just check` for SQLite + Postgres.
|
||||
|
||||
Run `just test-smoke` when you specifically need the MCP smoke flow.
|
||||
|
||||
If testmon is “cold,” the first run may be long. Subsequent runs get much faster.
|
||||
|
||||
### Consolidated Agent Package Checks
|
||||
|
||||
The monorepo ships several host-native packages alongside the Python core. Use the root justfile as the canonical entry point:
|
||||
|
||||
- `just package-check` — validates every copied package and generated bundle path.
|
||||
- `just package-check-claude-code` — validates the root and plugin-local Claude marketplace manifests, the SessionStart/PreCompact hooks, the bundled output style, and the seed schemas, then runs `claude plugin validate . --strict`.
|
||||
- `just package-check-skills` — validates every top-level `skills/memory-*/SKILL.md` frontmatter block.
|
||||
- `just package-check-hermes` — validates `integrations/hermes/plugin.yaml`, the Hermes provider entrypoint, bundled skill, and runs the hermetic unit suite.
|
||||
- `just package-check-openclaw` — runs the OpenClaw package install, copies top-level skills into the generated bundle, typechecks, lints, builds `dist/`, runs Bun tests, and performs `npm pack --dry-run`.
|
||||
- `just agent-harness-check` — checks the host-specific harnesses without the shared markdown-only skills target.
|
||||
|
||||
Package-local justfiles live in `plugins/claude-code/`, `skills/`, `integrations/hermes/`, and `integrations/openclaw/`. Prefer the root targets for PR verification so command names stay stable as package internals evolve.
|
||||
|
||||
### PR CI Gate
|
||||
|
||||
Before opening or updating a PR, run the checks that mirror the common required CI failures:
|
||||
|
||||
- Run `just typecheck` in addition to targeted `ruff` and `pytest` commands when tests were added or changed.
|
||||
- Sign commits with `git commit -s` so DCO passes. If a PR branch already has unsigned commits, rewrite the branch with signed-off commits before asking for review.
|
||||
- Use a semantic PR title accepted by `.github/workflows/pr-title.yml`: `type(scope): summary`.
|
||||
- Use one of the allowed scopes: `core`, `cli`, `api`, `mcp`, `sync`, `ui`, `ci`, `deps`, `installer`, `plugins`, `skills`, `integrations`.
|
||||
|
||||
### Test Structure
|
||||
|
||||
- `tests/` - Unit tests for individual components (mocked, fast)
|
||||
- `test-int/` - Integration tests for real-world scenarios (no mocks, realistic)
|
||||
- Both directories are covered by unified coverage reporting
|
||||
- Benchmark tests in `test-int/` are marked with `@pytest.mark.benchmark`
|
||||
- Slow tests are marked with `@pytest.mark.slow`
|
||||
- Smoke tests are marked with `@pytest.mark.smoke`
|
||||
|
||||
### Code Style Guidelines
|
||||
|
||||
- Line length: 100 characters max
|
||||
- Python 3.12+ with full type annotations (uses type parameters and type aliases)
|
||||
- Format with ruff (consistent styling)
|
||||
- Import order: standard lib, third-party, local imports
|
||||
- Naming: snake_case for functions/variables, PascalCase for classes
|
||||
- Prefer async patterns with SQLAlchemy 2.0
|
||||
- Use Pydantic v2 for data validation and schemas
|
||||
- CLI uses Typer for command structure
|
||||
- API uses FastAPI for endpoints
|
||||
- Follow the repository pattern for data access
|
||||
- Tools communicate to api routers via the httpx ASGI client (in process)
|
||||
|
||||
### Programming Style
|
||||
|
||||
See [docs/ENGINEERING_STYLE.md](docs/ENGINEERING_STYLE.md) for the fuller house style. The
|
||||
short version for agents:
|
||||
|
||||
- Prefer type-safe, explicit designs over object-heavy indirection. Use Python 3.12 `type`
|
||||
aliases, full annotations, and narrow `Protocol`s when a caller only needs a capability.
|
||||
- Use dataclasses for internal value objects and operation results; use Pydantic v2 at API,
|
||||
CLI, MCP, and persistence boundaries where validation and serialization matter.
|
||||
- Keep async boundaries obvious. Resource-owning code should use context managers, propagate
|
||||
cancellation, and avoid hidden background work unless the lifecycle is explicit.
|
||||
- Fail fast. Do not add silent fallback logic, broad exception swallowing, speculative
|
||||
`getattr`, or casts that hide an unclear model shape.
|
||||
- Keep control flow simple and local. Push branching decisions up, keep leaf helpers focused,
|
||||
and name values after the domain concept they carry.
|
||||
- Use evidence-first testing. Add or update meaningful regression tests for bugs and risky
|
||||
behavior, prefer real code paths over mocks, and run the narrowest command that proves the
|
||||
change before widening verification.
|
||||
- Comments should explain why a branch, invariant, or constraint exists. Avoid comments that
|
||||
merely narrate obvious code.
|
||||
|
||||
### Code Change Guidelines
|
||||
|
||||
- **Full file read before edits**: Before editing any file, read it in full first to ensure complete context; partial reads lead to corrupted edits
|
||||
- **Minimize diffs**: Prefer the smallest change that satisfies the request. Avoid unrelated refactors or style rewrites unless necessary for correctness
|
||||
- **House style is canonical**: Follow the Programming Style section above for type-safe,
|
||||
fail-fast code; do not hide unclear models with speculative attributes, broad exception
|
||||
handling, casts, or unapproved fallback logic
|
||||
- **No guessing**: Do not say "The issue is..." before you actually know what the issue is. Investigate first.
|
||||
|
||||
### Literate Programming Style
|
||||
|
||||
Code should tell a story. Comments must explain the "why" and narrative flow, not just the "what".
|
||||
|
||||
**Section Headers:**
|
||||
For files with multiple phases of logic, add section headers so the control flow reads like chapters:
|
||||
```python
|
||||
# --- Authentication ---
|
||||
# ... auth logic ...
|
||||
|
||||
# --- Data Validation ---
|
||||
# ... validation logic ...
|
||||
|
||||
# --- Business Logic ---
|
||||
# ... core logic ...
|
||||
```
|
||||
|
||||
**Decision Point Comments:**
|
||||
For conditionals that materially change behavior (gates, fallbacks, retries, feature flags), add comments with:
|
||||
- **Trigger**: what condition causes this branch
|
||||
- **Why**: the rationale (cost, correctness, UX, determinism)
|
||||
- **Outcome**: what changes downstream
|
||||
|
||||
```python
|
||||
# Trigger: project has no active sync watcher
|
||||
# Why: avoid duplicate file system watchers consuming resources
|
||||
# Outcome: starts new watcher, registers in active_watchers dict
|
||||
if project_id not in active_watchers:
|
||||
start_watcher(project_id)
|
||||
```
|
||||
|
||||
**Constraint Comments:**
|
||||
If code exists because of a constraint (async requirements, rate limits, schema compatibility), explain the constraint near the code:
|
||||
```python
|
||||
# SQLite requires WAL mode for concurrent read/write access
|
||||
connection.execute("PRAGMA journal_mode=WAL")
|
||||
```
|
||||
|
||||
**What NOT to Comment:**
|
||||
Avoid comments that restate obvious code:
|
||||
```python
|
||||
# Bad - restates code
|
||||
counter += 1 # increment counter
|
||||
|
||||
# Good - explains why
|
||||
counter += 1 # track retries for backoff calculation
|
||||
```
|
||||
|
||||
### Codebase Architecture
|
||||
|
||||
See [docs/ARCHITECTURE.md](docs/ARCHITECTURE.md) for detailed architecture documentation.
|
||||
|
||||
**Directory Structure:**
|
||||
- `/alembic` - Alembic db migrations
|
||||
- `/api` - FastAPI REST endpoints + `container.py` composition root
|
||||
- `/cli` - Typer CLI + `container.py` composition root
|
||||
- `/deps` - Feature-scoped FastAPI dependencies (config, db, projects, repositories, services, importers)
|
||||
- `/importers` - Import functionality for Claude, ChatGPT, and other sources
|
||||
- `/markdown` - Markdown parsing and processing
|
||||
- `/mcp` - MCP server + `container.py` composition root + `clients/` typed API clients
|
||||
- `/models` - SQLAlchemy ORM models
|
||||
- `/repository` - Data access layer
|
||||
- `/schemas` - Pydantic models for validation
|
||||
- `/services` - Business logic layer
|
||||
- `/sync` - File synchronization services + `coordinator.py` for lifecycle management
|
||||
- `/plugins/claude-code` - Claude Code plugin marketplace package, hooks, skills, and agent harness
|
||||
- `/skills` - Canonical framework-agnostic Basic Memory `SKILL.md` source
|
||||
- `/integrations/hermes` - Hermes memory-provider plugin
|
||||
- `/integrations/openclaw` - OpenClaw npm/TypeScript plugin
|
||||
|
||||
**Composition Roots:**
|
||||
Each entrypoint (API, MCP, CLI) has a composition root that:
|
||||
- Reads `ConfigManager` (the only place that reads global config)
|
||||
- Resolves runtime mode via `RuntimeMode` enum (TEST > CLOUD > LOCAL)
|
||||
- Provides dependencies to downstream code explicitly
|
||||
|
||||
**Typed API Clients (MCP):**
|
||||
MCP tools use typed clients in `mcp/clients/` to communicate with the API:
|
||||
- `KnowledgeClient` - Entity CRUD operations
|
||||
- `SearchClient` - Search operations
|
||||
- `MemoryClient` - Context building
|
||||
- `DirectoryClient` - Directory listing
|
||||
- `ResourceClient` - Resource reading
|
||||
- `ProjectClient` - Project management
|
||||
|
||||
Flow: MCP Tool → Typed Client → HTTP API → Router → Service → Repository
|
||||
|
||||
### Development Notes
|
||||
|
||||
- MCP tools are defined in src/basic_memory/mcp/tools/
|
||||
- MCP prompts are defined in src/basic_memory/mcp/prompts/
|
||||
- MCP tools should be atomic, composable operations
|
||||
- Use `textwrap.dedent()` for multi-line string formatting in prompts and tools
|
||||
- MCP Prompts are used to invoke tools and format content with instructions for an LLM
|
||||
- Schema changes require Alembic migrations
|
||||
- SQLite is used for indexing and full text search, files are source of truth
|
||||
- Testing uses pytest with asyncio support (strict mode)
|
||||
- Unit tests (`tests/`) use mocks when necessary; integration tests (`test-int/`) use real implementations
|
||||
- By default, tests run against SQLite (fast, no Docker needed)
|
||||
- Set `BASIC_MEMORY_TEST_POSTGRES=1` to run against Postgres (uses testcontainers - Docker required)
|
||||
- Each test runs in a standalone environment with isolated database and tmp_path directory
|
||||
- CI runs SQLite and Postgres tests in parallel for faster feedback
|
||||
- Performance benchmarks are in `test-int/test_sync_performance_benchmark.py`
|
||||
- Use pytest markers: `@pytest.mark.benchmark` for benchmarks, `@pytest.mark.slow` for slow tests
|
||||
- **Coverage must stay at 100%**: Write tests for new code. Only use `# pragma: no cover` when tests would require excessive mocking (e.g., TYPE_CHECKING blocks, error handlers that need failure injection, runtime-mode-dependent code paths)
|
||||
|
||||
### Async Client Pattern (Important!)
|
||||
|
||||
**MCP tools use `get_project_client()` for per-project routing:**
|
||||
|
||||
```python
|
||||
from basic_memory.mcp.project_context import get_project_client
|
||||
|
||||
@mcp.tool()
|
||||
async def my_tool(project: str | None = None, context: Context | None = None):
|
||||
async with get_project_client(project, context) as (client, active_project):
|
||||
# client is routed based on project's mode (local ASGI or cloud HTTP)
|
||||
response = await call_get(client, "/path")
|
||||
return response
|
||||
```
|
||||
|
||||
**CLI commands and non-project-scoped code use `get_client()` directly:**
|
||||
|
||||
```python
|
||||
from basic_memory.mcp.async_client import get_client
|
||||
|
||||
async def my_cli_command():
|
||||
async with get_client() as client:
|
||||
response = await call_get(client, "/path")
|
||||
return response
|
||||
|
||||
# Per-project routing (when project name is known):
|
||||
async with get_client(project_name="research") as client:
|
||||
...
|
||||
```
|
||||
|
||||
**Do NOT use:**
|
||||
- ❌ `from basic_memory.mcp.async_client import client` (deprecated module-level client)
|
||||
- ❌ Manual auth header management
|
||||
- ❌ `inject_auth_header()` (deleted)
|
||||
- ❌ Separate `get_client()` + `get_active_project()` in MCP tools (use `get_project_client()` instead)
|
||||
|
||||
**Key principles:**
|
||||
- Auth happens at client creation, not per-request
|
||||
- Proper resource management via context managers
|
||||
- Per-project routing: each project can be LOCAL or CLOUD independently
|
||||
- Cloud projects use API key (`cloud_api_key` in config) as Bearer token
|
||||
- Routing priority: factory injection > force-local > per-project cloud > global cloud > local ASGI
|
||||
- Factory pattern enables dependency injection for cloud consolidation
|
||||
|
||||
**For cloud app integration:**
|
||||
```python
|
||||
from basic_memory.mcp import async_client
|
||||
|
||||
# Set custom factory before importing tools
|
||||
async_client.set_client_factory(your_custom_factory)
|
||||
```
|
||||
|
||||
See SPEC-16 for full context manager refactor details.
|
||||
|
||||
### Release Process
|
||||
|
||||
Releases are driven by `just release` / `just beta` — never by a bare `git tag`. The recipes bump version metadata, run pre-flight checks, land the bump on `main` through a release PR, tag, and push the tag. GitHub Actions then publishes to PyPI and updates the Homebrew formula.
|
||||
|
||||
**Main requires PRs.** The `main` ruleset rejects direct pushes ("Changes must be made through a pull request") and the repo disallows merge commits, so the recipes push a `release/vX.Y.Z` branch, open a PR titled `chore(core): release vX.Y.Z`, rebase-merge it with `gh pr merge --rebase`, then tag the rebased bump commit on `main` (located by its commit subject, since rebasing rewrites the SHA) and push the tag. The CHANGELOG entry for the version must already be on `main` — land it via a normal PR before running the recipe (it pre-flight-checks for a `## vX.Y.Z` heading).
|
||||
|
||||
**Stable release:**
|
||||
|
||||
```
|
||||
just release v0.21.3
|
||||
```
|
||||
|
||||
The recipe runs `just lint` + `just typecheck`, then updates every release manifest through `scripts/update_versions.py`: `src/basic_memory/__init__.py`, `server.json`, the root Claude marketplace, the Claude Code plugin manifest and local marketplace, the Hermes `plugin.yaml`, and the OpenClaw `package.json`. It commits as `chore: update version to X.Y.Z for vX.Y.Z release` on a `release/vX.Y.Z` branch, lands it on `main` via a rebase-merged PR, then tags the rebased commit and pushes the tag. After the tag lands, the `Release` workflow builds the Python package, publishes to PyPI, creates the GitHub release with auto-generated notes, publishes the OpenClaw npm package, and updates the Homebrew formula. The recipe finishes by printing the post-release tasks the workflow doesn't cover.
|
||||
|
||||
**Beta release:** `just beta v0.21.3b1` — same flow with a beta-suffixed tag. PyPI consumers install with `pip install basic-memory --pre`.
|
||||
|
||||
**Release dry run:** `just release-dry-run v0.21.4` previews the consolidated version update without writing files.
|
||||
|
||||
**Development builds:** every commit to `main` publishes a `0.21.3.dev26+468a22f`-style version to PyPI automatically via `.github/workflows/dev-release.yml`. No human action.
|
||||
|
||||
**Do not tag releases by hand.** A bare `git tag vX.Y.Z` skips the in-code version bump. Package metadata is still correct (uv-dynamic-versioning derives it from the git tag) but `basic-memory --version` reports the previous release, which is what happened with v0.21.2 → v0.21.3.
|
||||
|
||||
**Post-release tasks** the recipe surfaces but doesn't run:
|
||||
- `docs.basicmemory.com` — add a What's New page under `content/2.whats-new/` and bump the version badge in `content/index.md` (the changelog page auto-fetches GitHub releases; see that repo's CLAUDE.md version-bump checklist)
|
||||
- `basicmemory.com` — the marketing site (Astro + React, repo
|
||||
`basicmachines-co/basicmemory.com`, formerly `basicmachines.co`) carries **no
|
||||
hardcoded version number** in its UI, so there is nothing to bump. For a
|
||||
significant release, optionally add a dated announcement post under
|
||||
`src/content/blog/` (model it on an existing `basic-memory-vX-Y-Z-release.md`).
|
||||
Skip entirely for routine patch releases.
|
||||
- MCP Registry — `mcp-publisher publish` from the repo root
|
||||
|
||||
See `.claude/commands/release/release.md` (and `beta.md`, `release-check.md`, `changelog.md` alongside it) for the full release + post-release runbook, including the slash commands.
|
||||
|
||||
## BASIC MEMORY PRODUCT USAGE
|
||||
|
||||
### Knowledge Structure
|
||||
|
||||
- Entity: Any concept, document, or idea represented as a markdown file
|
||||
- Observation: A categorized fact about an entity (`- [category] content`)
|
||||
- Relation: A directional link between entities (`- relation_type [[Target]]`)
|
||||
- Frontmatter: YAML metadata at the top of markdown files
|
||||
- Knowledge representation follows precise markdown format:
|
||||
- Observations with [category] prefixes
|
||||
- Relations with WikiLinks [[Entity]]
|
||||
- Frontmatter with metadata
|
||||
|
||||
### Basic Memory Commands
|
||||
|
||||
**Local Commands:**
|
||||
- Check sync status: `basic-memory status`
|
||||
- Doctor check (file <-> DB loop): `basic-memory doctor`
|
||||
- Import from Claude: `basic-memory import claude conversations`
|
||||
- Import from ChatGPT: `basic-memory import chatgpt`
|
||||
- Import from Memory JSON: `basic-memory import memory-json`
|
||||
- Tool access: `basic-memory tool` (provides CLI access to MCP tools)
|
||||
- Continue: `basic-memory tool continue-conversation --topic="search"`
|
||||
|
||||
**Project Management:**
|
||||
- List projects: `basic-memory project list`
|
||||
- Add project: `basic-memory project add "name" ~/path`
|
||||
- Project info: `basic-memory project info`
|
||||
- Set cloud mode: `basic-memory project set-cloud "name"`
|
||||
- Set local mode: `basic-memory project set-local "name"`
|
||||
- One-way sync (local -> cloud): `basic-memory project sync`
|
||||
- Bidirectional sync: `basic-memory project bisync`
|
||||
- Integrity check: `basic-memory project check`
|
||||
|
||||
**Cloud Commands (requires subscription):**
|
||||
- Authenticate (global): `basic-memory cloud login`
|
||||
- Logout (global): `basic-memory cloud logout`
|
||||
- Check cloud status: `basic-memory cloud status`
|
||||
- Setup cloud sync: `basic-memory cloud setup`
|
||||
- Save API key: `basic-memory cloud set-key bmc_...`
|
||||
- Create API key: `basic-memory cloud create-key "name"`
|
||||
- Manage snapshots: `basic-memory cloud snapshot [create|list|delete|show|browse]`
|
||||
- Restore from snapshot: `basic-memory cloud restore <path> --snapshot <id>`
|
||||
|
||||
**Cloud Sync Commands (Personal and Team workspaces):**
|
||||
- Fetch cloud changes (cloud -> local): `basic-memory cloud pull --name "name"` (Team-safe; additive, never deletes local)
|
||||
- Upload local changes (local -> cloud): `basic-memory cloud push --name "name"` (Team-safe; additive, never deletes cloud)
|
||||
- Resolve conflicts on push/pull: `--on-conflict [fail|keep-local|keep-cloud|keep-both]` (default `fail` lists conflicts and aborts, git-style)
|
||||
- One-way mirror (local -> cloud): `basic-memory cloud sync --name "name"` (Personal workspaces only; deletes cloud files missing locally)
|
||||
- Two-way mirror (local <-> cloud): `basic-memory cloud bisync --name "name"` (Personal workspaces only)
|
||||
|
||||
### MCP Capabilities
|
||||
|
||||
- Basic Memory exposes these MCP tools to LLMs:
|
||||
|
||||
**Content Management:**
|
||||
- `write_note(title, content, directory, tags)` - Create/update markdown notes with semantic observations and relations
|
||||
- `read_note(identifier, page, page_size)` - Read notes by title, permalink, or memory:// URL with knowledge graph awareness
|
||||
- `read_content(path)` - Read raw file content (text, images, binaries) without knowledge graph processing
|
||||
- `view_note(identifier, page, page_size)` - View notes as formatted artifacts for better readability
|
||||
- `edit_note(identifier, operation, content)` - Edit notes incrementally (append, prepend, find/replace, replace_section)
|
||||
- `move_note(identifier, destination_path, is_directory)` - Move notes or directories to new locations, updating database and maintaining links
|
||||
- `delete_note(identifier, is_directory)` - Delete notes or directories from the knowledge base
|
||||
|
||||
**Knowledge Graph Navigation:**
|
||||
- `build_context(url, depth, timeframe)` - Navigate the knowledge graph via memory:// URLs for conversation continuity
|
||||
- `recent_activity(type, depth, timeframe)` - Get recently updated information with specified timeframe (e.g., "1d", "1 week")
|
||||
- `list_directory(dir_name, depth, file_name_glob)` - Browse directory contents with filtering and depth control
|
||||
|
||||
**Search & Discovery:**
|
||||
- `search_notes(query, page, page_size, search_type, types, entity_types, after_date)` - Full-text search across all content with advanced filtering options
|
||||
|
||||
**Project Management:**
|
||||
- `list_memory_projects()` - List all available projects with their status
|
||||
- `create_memory_project(project_name, project_path, set_default)` - Create new Basic Memory projects
|
||||
- `delete_project(project_name)` - Delete a project from configuration
|
||||
|
||||
**Visualization:**
|
||||
- `canvas(nodes, edges, title, directory)` - Generate Obsidian canvas files for knowledge graph visualization
|
||||
|
||||
**ChatGPT-Compatible Tools:**
|
||||
- `search(query)` - Search across knowledge base (OpenAI actions compatible)
|
||||
- `fetch(id)` - Fetch full content of a search result document
|
||||
|
||||
- MCP Prompts for better AI interaction:
|
||||
- `ai_assistant_guide()` - Guidance on effectively using Basic Memory tools for AI assistants
|
||||
- `continue_conversation(topic, timeframe)` - Continue previous conversations with relevant historical context
|
||||
- `search(query, after_date)` - Search with detailed, formatted results for better context understanding
|
||||
- `recent_activity(timeframe)` - View recently changed items with formatted output
|
||||
|
||||
### Cloud Features (v0.15.0+)
|
||||
|
||||
Basic Memory now supports cloud synchronization and storage (requires active subscription):
|
||||
|
||||
**Authentication:**
|
||||
- JWT-based authentication with subscription validation
|
||||
- Secure session management with token refresh
|
||||
- Support for multiple cloud projects
|
||||
|
||||
**Bidirectional Sync:**
|
||||
- rclone bisync integration for two-way synchronization
|
||||
- Conflict resolution and integrity verification
|
||||
- Real-time sync with change detection
|
||||
- Mount/unmount cloud storage for direct file access
|
||||
|
||||
**Cloud Project Management:**
|
||||
- Create and manage projects in the cloud
|
||||
- Toggle between local and cloud modes
|
||||
- Per-project sync configuration
|
||||
- Subscription-based access control
|
||||
|
||||
**Security & Performance:**
|
||||
- Removed .env file loading for improved security
|
||||
- .gitignore integration (respects gitignored files)
|
||||
- WAL mode for SQLite performance
|
||||
- Background relation resolution (non-blocking startup)
|
||||
- API performance optimizations (SPEC-11)
|
||||
|
||||
**Per-Project Cloud Routing:**
|
||||
|
||||
Individual projects can be routed through the cloud while others stay local, using an API key:
|
||||
|
||||
```bash
|
||||
# Save API key and set project to cloud mode
|
||||
basic-memory cloud set-key bmc_abc123...
|
||||
basic-memory project set-cloud research # route through cloud
|
||||
basic-memory project set-local research # revert to local
|
||||
```
|
||||
|
||||
MCP tools use `get_project_client()` which automatically routes based on the project's mode. Cloud projects use the `cloud_api_key` from config as Bearer token.
|
||||
|
||||
**CLI Routing Flags (Global Cloud Mode):**
|
||||
|
||||
When global cloud mode is enabled, CLI commands route to the cloud API by default. Use `--local` and `--cloud` flags to override:
|
||||
|
||||
```bash
|
||||
# Force local routing (ignore cloud mode)
|
||||
basic-memory status --local
|
||||
basic-memory project list --local
|
||||
|
||||
# Force cloud routing (when cloud mode is disabled)
|
||||
basic-memory status --cloud
|
||||
basic-memory project info my-project --cloud
|
||||
```
|
||||
|
||||
Key behaviors:
|
||||
- The local MCP server (`basic-memory mcp`) automatically uses local routing
|
||||
- This allows simultaneous use of local Claude Desktop and cloud-based clients
|
||||
- Some commands (like `project default`, `project sync-config`, `project move`) require `--local` in cloud mode since they modify local configuration
|
||||
- Environment variable `BASIC_MEMORY_FORCE_LOCAL=true` forces local routing globally
|
||||
- Per-project cloud routing via API key works independently of global cloud mode
|
||||
|
||||
## AI-Human Collaborative Development
|
||||
|
||||
Basic Memory emerged from and enables a new kind of development process that combines human and AI capabilities. Instead
|
||||
of using AI just for code generation, we've developed a true collaborative workflow:
|
||||
|
||||
1. AI (LLM) writes initial implementation based on specifications and context
|
||||
2. Human reviews, runs tests, and commits code with any necessary adjustments
|
||||
3. Knowledge persists across conversations using Basic Memory's knowledge graph
|
||||
4. Development continues seamlessly across different AI sessions with consistent context
|
||||
5. Results improve through iterative collaboration and shared understanding
|
||||
|
||||
This approach has allowed us to tackle more complex challenges and build a more robust system than either humans or AI
|
||||
could achieve independently.
|
||||
|
||||
**Problem-Solving Guidance:**
|
||||
- If a solution isn't working after reasonable effort, suggest alternative approaches
|
||||
- Don't persist with a problematic library or pattern when better alternatives exist
|
||||
- Example: When py-pglite caused cascading test failures, switching to testcontainers-postgres was the right call
|
||||
|
||||
## GitHub Integration
|
||||
|
||||
Basic Memory has taken AI-Human collaboration to the next level by integrating Claude directly into the development workflow through GitHub:
|
||||
|
||||
### GitHub MCP Tools
|
||||
|
||||
Using the GitHub Model Context Protocol server, Claude can now:
|
||||
|
||||
- **Repository Management**:
|
||||
- View repository files and structure
|
||||
- Read file contents
|
||||
- Create new branches
|
||||
- Create and update files
|
||||
|
||||
- **Issue Management**:
|
||||
- Create new issues
|
||||
- Comment on existing issues
|
||||
- Close and update issues
|
||||
- Search across issues
|
||||
|
||||
- **Pull Request Workflow**:
|
||||
- Create pull requests
|
||||
- Review code changes
|
||||
- Add comments to PRs
|
||||
|
||||
This integration enables Claude to participate as a full team member in the development process, not just as a code generation tool. Claude's GitHub account ([bm-claudeai](https://github.com/bm-claudeai)) is a member of the Basic Machines organization with direct contributor access to the codebase.
|
||||
|
||||
### Collaborative Development Process
|
||||
|
||||
With GitHub integration, the development workflow includes:
|
||||
|
||||
1. **Direct code review** - Claude can analyze PRs and provide detailed feedback
|
||||
2. **Contribution tracking** - All of Claude's contributions are properly attributed in the Git history
|
||||
3. **Branch management** - Claude can create feature branches for implementations
|
||||
4. **Documentation maintenance** - Claude can keep documentation updated as the code evolves
|
||||
5. **Code Commits**: ALWAYS sign off commits with `git commit -s`
|
||||
6. **Pull Request Titles**: PR titles must follow the semantic format enforced by `.github/workflows/pr-title.yml`: `type(scope): summary`
|
||||
- Allowed types: `feat`, `fix`, `chore`, `docs`, `style`, `refactor`, `perf`, `test`, `build`, `ci`
|
||||
- Allowed scopes: `core`, `cli`, `api`, `mcp`, `sync`, `ui`, `ci`, `deps`, `installer`, `plugins`, `skills`, `integrations`
|
||||
- Example: `fix(cli): propagate cloud workspace routing`
|
||||
|
||||
This level of integration represents a new paradigm in AI-human collaboration, where the AI assistant becomes a full-fledged team member rather than just a tool for generating code snippets.
|
||||
+5
-967
File diff suppressed because it is too large
Load Diff
@@ -0,0 +1,268 @@
|
||||
# CLAUDE.md - Basic Memory Project Guide
|
||||
|
||||
## Project Overview
|
||||
|
||||
Basic Memory is a local-first knowledge management system built on the Model Context Protocol (MCP). It enables
|
||||
bidirectional communication between LLMs (like Claude) and markdown files, creating a personal knowledge graph that can
|
||||
be traversed using links between documents.
|
||||
|
||||
## CODEBASE DEVELOPMENT
|
||||
|
||||
### Project information
|
||||
|
||||
See the [README.md](README.md) file for a project overview.
|
||||
|
||||
### Build and Test Commands
|
||||
|
||||
- Install: `just install` or `pip install -e ".[dev]"`
|
||||
- Run all tests (with coverage): `just test` - Runs both unit and integration tests with unified coverage
|
||||
- Run unit tests only: `just test-unit` - Fast, no coverage
|
||||
- Run integration tests only: `just test-int` - Fast, no coverage
|
||||
- Generate HTML coverage: `just coverage` - Opens in browser
|
||||
- Single test: `pytest tests/path/to/test_file.py::test_function_name`
|
||||
- Run benchmarks: `pytest test-int/test_sync_performance_benchmark.py -v -m "benchmark and not slow"`
|
||||
- Lint: `just lint` or `ruff check . --fix`
|
||||
- Type check: `just typecheck` or `uv run pyright`
|
||||
- Format: `just format` or `uv run ruff format .`
|
||||
- Run all code checks: `just check` (runs lint, format, typecheck, test)
|
||||
- Create db migration: `just migration "Your migration message"`
|
||||
- Run development MCP Inspector: `just run-inspector`
|
||||
|
||||
**Note:** Project requires Python 3.12+ (uses type parameter syntax and `type` aliases introduced in 3.12)
|
||||
|
||||
### Test Structure
|
||||
|
||||
- `tests/` - Unit tests for individual components (mocked, fast)
|
||||
- `test-int/` - Integration tests for real-world scenarios (no mocks, realistic)
|
||||
- Both directories are covered by unified coverage reporting
|
||||
- Benchmark tests in `test-int/` are marked with `@pytest.mark.benchmark`
|
||||
- Slow tests are marked with `@pytest.mark.slow`
|
||||
|
||||
### Code Style Guidelines
|
||||
|
||||
- Line length: 100 characters max
|
||||
- Python 3.12+ with full type annotations (uses type parameters and type aliases)
|
||||
- Format with ruff (consistent styling)
|
||||
- Import order: standard lib, third-party, local imports
|
||||
- Naming: snake_case for functions/variables, PascalCase for classes
|
||||
- Prefer async patterns with SQLAlchemy 2.0
|
||||
- Use Pydantic v2 for data validation and schemas
|
||||
- CLI uses Typer for command structure
|
||||
- API uses FastAPI for endpoints
|
||||
- Follow the repository pattern for data access
|
||||
- Tools communicate to api routers via the httpx ASGI client (in process)
|
||||
|
||||
### Codebase Architecture
|
||||
|
||||
- `/alembic` - Alembic db migrations
|
||||
- `/api` - FastAPI implementation of REST endpoints
|
||||
- `/cli` - Typer command-line interface
|
||||
- `/markdown` - Markdown parsing and processing
|
||||
- `/mcp` - Model Context Protocol server implementation
|
||||
- `/models` - SQLAlchemy ORM models
|
||||
- `/repository` - Data access layer
|
||||
- `/schemas` - Pydantic models for validation
|
||||
- `/services` - Business logic layer
|
||||
- `/sync` - File synchronization services
|
||||
|
||||
### Development Notes
|
||||
|
||||
- MCP tools are defined in src/basic_memory/mcp/tools/
|
||||
- MCP prompts are defined in src/basic_memory/mcp/prompts/
|
||||
- MCP tools should be atomic, composable operations
|
||||
- Use `textwrap.dedent()` for multi-line string formatting in prompts and tools
|
||||
- MCP Prompts are used to invoke tools and format content with instructions for an LLM
|
||||
- Schema changes require Alembic migrations
|
||||
- SQLite is used for indexing and full text search, files are source of truth
|
||||
- Testing uses pytest with asyncio support (strict mode)
|
||||
- Unit tests (`tests/`) use mocks when necessary; integration tests (`test-int/`) use real implementations
|
||||
- Test database uses in-memory SQLite
|
||||
- Each test runs in a standalone environment with in-memory SQLite and tmp_file directory
|
||||
- Performance benchmarks are in `test-int/test_sync_performance_benchmark.py`
|
||||
- Use pytest markers: `@pytest.mark.benchmark` for benchmarks, `@pytest.mark.slow` for slow tests
|
||||
|
||||
### Async Client Pattern (Important!)
|
||||
|
||||
**All MCP tools and CLI commands use the context manager pattern for HTTP clients:**
|
||||
|
||||
```python
|
||||
from basic_memory.mcp.async_client import get_client
|
||||
|
||||
async def my_mcp_tool():
|
||||
async with get_client() as client:
|
||||
# Use client for API calls
|
||||
response = await call_get(client, "/path")
|
||||
return response
|
||||
```
|
||||
|
||||
**Do NOT use:**
|
||||
- ❌ `from basic_memory.mcp.async_client import client` (deprecated module-level client)
|
||||
- ❌ Manual auth header management
|
||||
- ❌ `inject_auth_header()` (deleted)
|
||||
|
||||
**Key principles:**
|
||||
- Auth happens at client creation, not per-request
|
||||
- Proper resource management via context managers
|
||||
- Supports three modes: Local (ASGI), CLI cloud (HTTP + auth), Cloud app (factory injection)
|
||||
- Factory pattern enables dependency injection for cloud consolidation
|
||||
|
||||
**For cloud app integration:**
|
||||
```python
|
||||
from basic_memory.mcp import async_client
|
||||
|
||||
# Set custom factory before importing tools
|
||||
async_client.set_client_factory(your_custom_factory)
|
||||
```
|
||||
|
||||
See SPEC-16 for full context manager refactor details.
|
||||
|
||||
## BASIC MEMORY PRODUCT USAGE
|
||||
|
||||
### Knowledge Structure
|
||||
|
||||
- Entity: Any concept, document, or idea represented as a markdown file
|
||||
- Observation: A categorized fact about an entity (`- [category] content`)
|
||||
- Relation: A directional link between entities (`- relation_type [[Target]]`)
|
||||
- Frontmatter: YAML metadata at the top of markdown files
|
||||
- Knowledge representation follows precise markdown format:
|
||||
- Observations with [category] prefixes
|
||||
- Relations with WikiLinks [[Entity]]
|
||||
- Frontmatter with metadata
|
||||
|
||||
### Basic Memory Commands
|
||||
|
||||
**Local Commands:**
|
||||
- Sync knowledge: `basic-memory sync` or `basic-memory sync --watch`
|
||||
- Import from Claude: `basic-memory import claude conversations`
|
||||
- Import from ChatGPT: `basic-memory import chatgpt`
|
||||
- Import from Memory JSON: `basic-memory import memory-json`
|
||||
- Check sync status: `basic-memory status`
|
||||
- Tool access: `basic-memory tools` (provides CLI access to MCP tools)
|
||||
- Guide: `basic-memory tools basic-memory-guide`
|
||||
- Continue: `basic-memory tools continue-conversation --topic="search"`
|
||||
|
||||
**Cloud Commands (requires subscription):**
|
||||
- Authenticate: `basic-memory cloud login`
|
||||
- Logout: `basic-memory cloud logout`
|
||||
- Bidirectional sync: `basic-memory cloud sync`
|
||||
- Integrity check: `basic-memory cloud check`
|
||||
- Mount cloud storage: `basic-memory cloud mount`
|
||||
- Unmount cloud storage: `basic-memory cloud unmount`
|
||||
|
||||
### MCP Capabilities
|
||||
|
||||
- Basic Memory exposes these MCP tools to LLMs:
|
||||
|
||||
**Content Management:**
|
||||
- `write_note(title, content, folder, tags)` - Create/update markdown notes with semantic observations and relations
|
||||
- `read_note(identifier, page, page_size)` - Read notes by title, permalink, or memory:// URL with knowledge graph awareness
|
||||
- `read_content(path)` - Read raw file content (text, images, binaries) without knowledge graph processing
|
||||
- `view_note(identifier, page, page_size)` - View notes as formatted artifacts for better readability
|
||||
- `edit_note(identifier, operation, content)` - Edit notes incrementally (append, prepend, find/replace, replace_section)
|
||||
- `move_note(identifier, destination_path)` - Move notes to new locations, updating database and maintaining links
|
||||
- `delete_note(identifier)` - Delete notes from the knowledge base
|
||||
|
||||
**Knowledge Graph Navigation:**
|
||||
- `build_context(url, depth, timeframe)` - Navigate the knowledge graph via memory:// URLs for conversation continuity
|
||||
- `recent_activity(type, depth, timeframe)` - Get recently updated information with specified timeframe (e.g., "1d", "1 week")
|
||||
- `list_directory(dir_name, depth, file_name_glob)` - Browse directory contents with filtering and depth control
|
||||
|
||||
**Search & Discovery:**
|
||||
- `search_notes(query, page, page_size, search_type, types, entity_types, after_date)` - Full-text search across all content with advanced filtering options
|
||||
|
||||
**Project Management:**
|
||||
- `list_memory_projects()` - List all available projects with their status
|
||||
- `create_memory_project(project_name, project_path, set_default)` - Create new Basic Memory projects
|
||||
- `delete_project(project_name)` - Delete a project from configuration
|
||||
- `get_current_project()` - Get current project information and stats
|
||||
- `sync_status()` - Check file synchronization and background operation status
|
||||
|
||||
**Visualization:**
|
||||
- `canvas(nodes, edges, title, folder)` - Generate Obsidian canvas files for knowledge graph visualization
|
||||
|
||||
- MCP Prompts for better AI interaction:
|
||||
- `ai_assistant_guide()` - Guidance on effectively using Basic Memory tools for AI assistants
|
||||
- `continue_conversation(topic, timeframe)` - Continue previous conversations with relevant historical context
|
||||
- `search(query, after_date)` - Search with detailed, formatted results for better context understanding
|
||||
- `recent_activity(timeframe)` - View recently changed items with formatted output
|
||||
- `json_canvas_spec()` - Full JSON Canvas specification for Obsidian visualization
|
||||
|
||||
### Cloud Features (v0.15.0+)
|
||||
|
||||
Basic Memory now supports cloud synchronization and storage (requires active subscription):
|
||||
|
||||
**Authentication:**
|
||||
- JWT-based authentication with subscription validation
|
||||
- Secure session management with token refresh
|
||||
- Support for multiple cloud projects
|
||||
|
||||
**Bidirectional Sync:**
|
||||
- rclone bisync integration for two-way synchronization
|
||||
- Conflict resolution and integrity verification
|
||||
- Real-time sync with change detection
|
||||
- Mount/unmount cloud storage for direct file access
|
||||
|
||||
**Cloud Project Management:**
|
||||
- Create and manage projects in the cloud
|
||||
- Toggle between local and cloud modes
|
||||
- Per-project sync configuration
|
||||
- Subscription-based access control
|
||||
|
||||
**Security & Performance:**
|
||||
- Removed .env file loading for improved security
|
||||
- .gitignore integration (respects gitignored files)
|
||||
- WAL mode for SQLite performance
|
||||
- Background relation resolution (non-blocking startup)
|
||||
- API performance optimizations (SPEC-11)
|
||||
|
||||
## AI-Human Collaborative Development
|
||||
|
||||
Basic Memory emerged from and enables a new kind of development process that combines human and AI capabilities. Instead
|
||||
of using AI just for code generation, we've developed a true collaborative workflow:
|
||||
|
||||
1. AI (LLM) writes initial implementation based on specifications and context
|
||||
2. Human reviews, runs tests, and commits code with any necessary adjustments
|
||||
3. Knowledge persists across conversations using Basic Memory's knowledge graph
|
||||
4. Development continues seamlessly across different AI sessions with consistent context
|
||||
5. Results improve through iterative collaboration and shared understanding
|
||||
|
||||
This approach has allowed us to tackle more complex challenges and build a more robust system than either humans or AI
|
||||
could achieve independently.
|
||||
|
||||
## GitHub Integration
|
||||
|
||||
Basic Memory has taken AI-Human collaboration to the next level by integrating Claude directly into the development workflow through GitHub:
|
||||
|
||||
### GitHub MCP Tools
|
||||
|
||||
Using the GitHub Model Context Protocol server, Claude can now:
|
||||
|
||||
- **Repository Management**:
|
||||
- View repository files and structure
|
||||
- Read file contents
|
||||
- Create new branches
|
||||
- Create and update files
|
||||
|
||||
- **Issue Management**:
|
||||
- Create new issues
|
||||
- Comment on existing issues
|
||||
- Close and update issues
|
||||
- Search across issues
|
||||
|
||||
- **Pull Request Workflow**:
|
||||
- Create pull requests
|
||||
- Review code changes
|
||||
- Add comments to PRs
|
||||
|
||||
This integration enables Claude to participate as a full team member in the development process, not just as a code generation tool. Claude's GitHub account ([bm-claudeai](https://github.com/bm-claudeai)) is a member of the Basic Machines organization with direct contributor access to the codebase.
|
||||
|
||||
### Collaborative Development Process
|
||||
|
||||
With GitHub integration, the development workflow includes:
|
||||
|
||||
1. **Direct code review** - Claude can analyze PRs and provide detailed feedback
|
||||
2. **Contribution tracking** - All of Claude's contributions are properly attributed in the Git history
|
||||
3. **Branch management** - Claude can create feature branches for implementations
|
||||
4. **Documentation maintenance** - Claude can keep documentation updated as the code evolves
|
||||
|
||||
This level of integration represents a new paradigm in AI-human collaboration, where the AI assistant becomes a full-fledged team member rather than just a tool for generating code snippets.
|
||||
@@ -224,6 +224,40 @@ See `test-int/BENCHMARKS.md` for detailed benchmark documentation.
|
||||
- **Fixtures**: Use async pytest fixtures for setup and teardown
|
||||
- **Markers**: Use `@pytest.mark.benchmark` for benchmarks, `@pytest.mark.slow` for slow tests
|
||||
|
||||
## Release Process
|
||||
|
||||
Basic Memory uses automatic versioning based on git tags with `uv-dynamic-versioning`. Here's how releases work:
|
||||
|
||||
### Version Management
|
||||
- **Development versions**: Automatically generated from git commits (e.g., `0.12.4.dev26+468a22f`)
|
||||
- **Beta releases**: Created by tagging with beta suffixes (e.g., `git tag v0.13.0b1`)
|
||||
- **Stable releases**: Created by tagging with version numbers (e.g., `git tag v0.13.0`)
|
||||
|
||||
### Release Workflows
|
||||
|
||||
#### Development Builds
|
||||
- Automatically published to PyPI on every commit to `main`
|
||||
- Version format: `0.12.4.dev26+468a22f` (base version + dev + commit count + hash)
|
||||
- Users install with: `pip install basic-memory --pre --force-reinstall`
|
||||
|
||||
#### Beta Releases
|
||||
1. Create and push a beta tag: `git tag v0.13.0b1 && git push origin v0.13.0b1`
|
||||
2. GitHub Actions automatically builds and publishes to PyPI
|
||||
3. Users install with: `pip install basic-memory --pre`
|
||||
|
||||
#### Stable Releases
|
||||
1. Create and push a version tag: `git tag v0.13.0 && git push origin v0.13.0`
|
||||
2. GitHub Actions automatically:
|
||||
- Builds the package with version `0.13.0`
|
||||
- Creates GitHub release with auto-generated notes
|
||||
- Publishes to PyPI
|
||||
3. Users install with: `pip install basic-memory`
|
||||
|
||||
### For Contributors
|
||||
- No manual version bumping required
|
||||
- Versions are automatically derived from git tags
|
||||
- Focus on code changes, not version management
|
||||
|
||||
## Creating Issues
|
||||
|
||||
If you're planning to work on something, please create an issue first to discuss the approach. Include:
|
||||
|
||||
+4
-10
@@ -8,13 +8,8 @@ ARG GID=1000
|
||||
COPY --from=ghcr.io/astral-sh/uv:latest /uv /uvx /bin/
|
||||
|
||||
# Set environment variables
|
||||
# UV_PYTHON_INSTALL_DIR ensures Python is installed to a persistent location
|
||||
# that survives in the final image (not in /root/.local which gets lost)
|
||||
# UV_PYTHON_PREFERENCE=only-managed tells uv to use its managed Python version
|
||||
ENV PYTHONUNBUFFERED=1 \
|
||||
PYTHONDONTWRITEBYTECODE=1 \
|
||||
UV_PYTHON_INSTALL_DIR=/python \
|
||||
UV_PYTHON_PREFERENCE=only-managed
|
||||
PYTHONDONTWRITEBYTECODE=1
|
||||
|
||||
# Create a group and user with the provided UID/GID
|
||||
# Check if the GID already exists, if not create appgroup
|
||||
@@ -24,10 +19,9 @@ RUN (getent group ${GID} || groupadd --gid ${GID} appgroup) && \
|
||||
# Copy the project into the image
|
||||
ADD . /app
|
||||
|
||||
# Install Python 3.13 explicitly and sync the project
|
||||
# Sync the project into a new environment, asserting the lockfile is up to date
|
||||
WORKDIR /app
|
||||
RUN uv python install 3.13
|
||||
RUN uv sync --locked --python 3.13
|
||||
RUN uv sync --locked
|
||||
|
||||
# Create necessary directories and set ownership
|
||||
RUN mkdir -p /app/data/basic-memory /app/.basic-memory && \
|
||||
@@ -49,4 +43,4 @@ HEALTHCHECK --interval=30s --timeout=10s --start-period=5s --retries=3 \
|
||||
CMD basic-memory --version || exit 1
|
||||
|
||||
# Use the basic-memory entrypoint to run the MCP server with default SSE transport
|
||||
CMD ["basic-memory", "mcp", "--transport", "sse", "--host", "0.0.0.0", "--port", "8000"]
|
||||
CMD ["basic-memory", "mcp", "--transport", "sse", "--host", "0.0.0.0", "--port", "8000"]
|
||||
-494
@@ -1,494 +0,0 @@
|
||||
# Note Format Reference
|
||||
|
||||
Every document in Basic Memory is a plain Markdown file. Files are the source of truth — changes to files automatically update the knowledge graph in the database. You maintain complete ownership, files work with git, and knowledge persists independently of any AI conversation.
|
||||
|
||||
## Document Structure
|
||||
|
||||
A note has three parts: YAML frontmatter, content (observations), and relations.
|
||||
|
||||
```markdown
|
||||
---
|
||||
title: Coffee Brewing Methods
|
||||
type: note
|
||||
tags: [coffee, brewing]
|
||||
permalink: coffee-brewing-methods
|
||||
---
|
||||
|
||||
# Coffee Brewing Methods
|
||||
|
||||
## Observations
|
||||
- [method] Pour over provides more flavor clarity than French press
|
||||
- [technique] Water temperature at 205°F extracts optimal compounds #brewing
|
||||
- [preference] Ethiopian beans work well with lighter roasts (personal experience)
|
||||
|
||||
## Relations
|
||||
- relates_to [[Coffee Bean Origins]]
|
||||
- requires [[Proper Grinding Technique]]
|
||||
- contrasts_with [[Tea Brewing Methods]]
|
||||
```
|
||||
|
||||
The `## Observations` and `## Relations` headings are conventional but not required — the parser detects observations and relations by their syntax patterns anywhere in the document.
|
||||
|
||||
## Frontmatter
|
||||
|
||||
YAML metadata between `---` fences at the top of the file.
|
||||
|
||||
| Field | Required | Default | Description |
|
||||
|-------|----------|---------|-------------|
|
||||
| `title` | No | filename stem | Used for linking and references. Auto-set from filename if missing. |
|
||||
| `type` | No | `note` | Entity type. Used for schema resolution and filtering. |
|
||||
| `tags` | No | `[]` | List or comma-separated string. Used for organization and search. |
|
||||
| `permalink` | No | generated from title | Stable identifier. Persists even if the file moves. |
|
||||
| `schema` | No | none | Schema attachment — dict (inline), string (reference), or omitted (implicit). |
|
||||
|
||||
Custom fields are allowed. Any key not in the standard set is stored as `entity_metadata` and indexed for search and filtering.
|
||||
|
||||
```yaml
|
||||
---
|
||||
title: Paul Graham
|
||||
type: Person
|
||||
tags: [startups, essays, lisp]
|
||||
permalink: paul-graham
|
||||
status: active
|
||||
source: wikipedia
|
||||
---
|
||||
```
|
||||
|
||||
Here `status` and `source` are custom fields stored in `entity_metadata`.
|
||||
|
||||
### Frontmatter Value Handling
|
||||
|
||||
YAML automatically converts some values to native types. Basic Memory normalizes them:
|
||||
|
||||
- Date strings (`2025-10-24`) → kept as ISO format strings
|
||||
- Numbers (`1.0`) → converted to strings
|
||||
- Booleans (`true`) → converted to strings (`"True"`)
|
||||
- Lists and dicts → preserved, items normalized recursively
|
||||
|
||||
This prevents errors when downstream code expects string values.
|
||||
|
||||
## Observations
|
||||
|
||||
An observation is a categorized fact about the entity. Written as a Markdown list item.
|
||||
|
||||
**Syntax:**
|
||||
|
||||
```
|
||||
- [category] content text #tag1 #tag2 (context)
|
||||
```
|
||||
|
||||
| Part | Required | Description |
|
||||
|------|----------|-------------|
|
||||
| `[category]` | Yes | Classification in square brackets. Any text except `[]()` chars. |
|
||||
| content | Yes | The fact or statement. |
|
||||
| `#tags` | No | Inline tags. Space-separated, each starting with `#`. |
|
||||
| `(context)` | No | Parenthesized text at end of line. Supporting details or source. |
|
||||
|
||||
### Examples
|
||||
|
||||
```markdown
|
||||
- [tech] Uses SQLite for storage #database
|
||||
- [design] Follows local-first architecture #architecture
|
||||
- [decision] Selected bcrypt for passwords #security (based on OWASP audit)
|
||||
- [name] Paul Graham
|
||||
- [expertise] Startups
|
||||
- [expertise] Lisp
|
||||
- [expertise] Essay writing
|
||||
```
|
||||
|
||||
Array-like fields use repeated categories — multiple `[expertise]` observations above.
|
||||
|
||||
### What Is Not an Observation
|
||||
|
||||
The parser excludes these list item patterns:
|
||||
|
||||
| Pattern | Example | Reason |
|
||||
|---------|---------|--------|
|
||||
| Checkboxes | `- [ ] Todo item`, `- [x] Done`, `- [-] Cancelled` | Task list syntax |
|
||||
| Markdown links | `- [text](url)` | URL link syntax |
|
||||
| Bare wiki links | `- [[Target]]` | Treated as a relation instead |
|
||||
|
||||
A list item with `#tags` but no `[category]` is still parsed — the tags are extracted and the category defaults to `Note`.
|
||||
|
||||
## Relations
|
||||
|
||||
Relations connect documents to form the knowledge graph. There are two kinds.
|
||||
|
||||
### Explicit Relations
|
||||
|
||||
Written as list items with a relation type and a `[[wiki link]]` target.
|
||||
|
||||
**Syntax:**
|
||||
|
||||
```
|
||||
- relation_type [[Target Entity]] (context)
|
||||
```
|
||||
|
||||
| Part | Required | Description |
|
||||
|------|----------|-------------|
|
||||
| `relation_type` | No | Text before `[[`. Defaults to `relates_to` if omitted. |
|
||||
| `[[Target]]` | Yes | Wiki link to the target entity. Matched by title or permalink. |
|
||||
| `(context)` | No | Parenthesized text after `]]`. Supporting details. |
|
||||
|
||||
### Examples
|
||||
|
||||
```markdown
|
||||
- implements [[Search Design]]
|
||||
- depends_on [[Database Schema]]
|
||||
- works_at [[Y Combinator]] (co-founder)
|
||||
- [[Some Entity]]
|
||||
```
|
||||
|
||||
The last example — a bare `[[wiki link]]` in a list item — gets relation type `relates_to`.
|
||||
|
||||
Common relation types:
|
||||
- `implements`, `depends_on`, `relates_to`, `inspired_by`
|
||||
- `extends`, `part_of`, `contains`, `pairs_with`
|
||||
- `works_at`, `authored`, `collaborated_with`
|
||||
|
||||
Any text works as a relation type. These are conventions, not a fixed set.
|
||||
|
||||
### Inline References
|
||||
|
||||
Wiki links appearing in regular prose (not as list items) create implicit `links_to` relations.
|
||||
|
||||
```markdown
|
||||
This builds on [[Core Design]] and uses [[Utility Functions]].
|
||||
```
|
||||
|
||||
This creates two relations: `links_to [[Core Design]]` and `links_to [[Utility Functions]]`.
|
||||
|
||||
### Forward References
|
||||
|
||||
Relations can link to entities that don't exist yet. Basic Memory resolves them when the target is created.
|
||||
|
||||
## Permalinks and memory:// URLs
|
||||
|
||||
Every document has a unique **permalink** — a stable identifier derived from its title. You can set one explicitly in frontmatter, or let the system generate it.
|
||||
|
||||
```yaml
|
||||
permalink: auth-approaches-2024
|
||||
```
|
||||
|
||||
Permalinks form the basis of `memory://` URLs:
|
||||
|
||||
```
|
||||
memory://auth-approaches-2024 # By permalink
|
||||
memory://Authentication Approaches # By title (auto-resolves)
|
||||
memory://project/auth-approaches # By path
|
||||
```
|
||||
|
||||
Pattern matching is supported:
|
||||
|
||||
```
|
||||
memory://auth* # Starts with "auth"
|
||||
memory://*/approaches # Ends with "approaches"
|
||||
memory://project/*/requirements # Nested wildcard
|
||||
```
|
||||
|
||||
## Schemas
|
||||
|
||||
Schemas declare the expected structure of a note — which observation categories and relation types a well-formed note should have. They use Picoschema, a compact notation from Google's Dotprompt that fits naturally in YAML frontmatter.
|
||||
|
||||
### Picoschema Syntax
|
||||
|
||||
```yaml
|
||||
schema:
|
||||
name: string, full name # required field with description
|
||||
email?: string, contact email # ? = optional
|
||||
role?: string, job title
|
||||
works_at?: Organization, employer # capitalized type = entity reference
|
||||
tags?(array): string, categories # array of type
|
||||
status?(enum): [active, inactive] # enum with allowed values
|
||||
metadata?(object): # nested object
|
||||
updated_at?: string
|
||||
source?: string
|
||||
```
|
||||
|
||||
| Notation | Meaning | Example |
|
||||
|----------|---------|---------|
|
||||
| `field: type` | Required field | `name: string` |
|
||||
| `field?: type` | Optional field | `role?: string` |
|
||||
| `field(array): type` | Array of values | `expertise(array): string` |
|
||||
| `field?(enum): [vals]` | Enum with allowed values | `status?(enum): [active, inactive]` |
|
||||
| `field?(object):` | Nested object with sub-fields | `metadata?(object):` |
|
||||
| `, description` | Description after comma | `name: string, full name` |
|
||||
| `EntityName` | Capitalized type = entity reference | `works_at?: Organization` |
|
||||
|
||||
**Scalar types:** `string`, `integer`, `number`, `boolean`, `any`
|
||||
|
||||
Any type not in that set whose first letter is uppercase is treated as an entity reference (a relation target).
|
||||
|
||||
### Schema-to-Note Mapping
|
||||
|
||||
Schemas validate against existing observation/relation syntax. Note authors don't learn new syntax.
|
||||
|
||||
| Schema Declaration | Maps To | Example in Note |
|
||||
|--------------------|---------|-----------------|
|
||||
| `field: string` | Observation `[field] value` | `- [name] Paul Graham` |
|
||||
| `field?(array): string` | Multiple `[field]` observations | `- [expertise] Lisp` (repeated) |
|
||||
| `field?: EntityType` | Relation `field [[Target]]` | `- works_at [[Y Combinator]]` |
|
||||
| `field?(array): EntityType` | Multiple `field` relations | `- authored [[Book]]` (repeated) |
|
||||
| `tags` | Frontmatter `tags` array | `tags: [startups, essays]` |
|
||||
| `field?(enum): [vals]` | Observation `[field] value` where value is in the set | `- [status] active` |
|
||||
|
||||
Observations and relations not covered by the schema are valid — schemas describe a subset, not a straitjacket.
|
||||
|
||||
### Schema Attachment
|
||||
|
||||
Three ways to attach a schema to a note, resolved in priority order:
|
||||
|
||||
**1. Inline schema** — `schema` is a dict in frontmatter:
|
||||
|
||||
```yaml
|
||||
---
|
||||
title: Team Standup 2024-01-15
|
||||
type: meeting
|
||||
schema:
|
||||
attendees(array): string, who was there
|
||||
decisions(array): string, what was decided
|
||||
action_items(array): string, follow-ups
|
||||
blockers?(array): string, anything stuck
|
||||
---
|
||||
```
|
||||
|
||||
Good for one-off structured notes or prototyping a schema before extracting it.
|
||||
|
||||
**2. Explicit reference** — `schema` is a string naming a schema note:
|
||||
|
||||
```yaml
|
||||
---
|
||||
title: Basic Memory
|
||||
schema: SoftwareProject
|
||||
---
|
||||
```
|
||||
|
||||
or by permalink:
|
||||
|
||||
```yaml
|
||||
---
|
||||
title: LLM Memory Patterns
|
||||
schema: schema/research-project
|
||||
---
|
||||
```
|
||||
|
||||
Use when the note's `type` differs from the schema it should validate against, or when multiple schema variants exist.
|
||||
|
||||
**3. Implicit by type** — no `schema` field, resolved by matching `type`:
|
||||
|
||||
```yaml
|
||||
---
|
||||
title: Paul Graham
|
||||
type: Person
|
||||
---
|
||||
```
|
||||
|
||||
The system looks up a schema note where `entity: Person`. If found, it applies. If not, no validation occurs.
|
||||
|
||||
**4. No schema** — perfectly fine. Most notes don't need one.
|
||||
|
||||
### Schema Notes
|
||||
|
||||
A schema is itself a Basic Memory note with `type: schema`. It lives anywhere (though `schema/` is the conventional directory).
|
||||
|
||||
```yaml
|
||||
# schema/Person.md
|
||||
---
|
||||
title: Person
|
||||
type: schema
|
||||
entity: Person
|
||||
version: 1
|
||||
schema:
|
||||
name: string, full name
|
||||
role?: string, job title or position
|
||||
works_at?: Organization, employer
|
||||
expertise?(array): string, areas of knowledge
|
||||
email?: string, contact email
|
||||
settings:
|
||||
validation: warn
|
||||
---
|
||||
|
||||
# Person
|
||||
|
||||
A human individual in the knowledge graph.
|
||||
```
|
||||
|
||||
| Field | Required | Description |
|
||||
|-------|----------|-------------|
|
||||
| `type` | Yes | Must be `schema` |
|
||||
| `entity` | Yes | The entity type this schema describes (e.g., `Person`) |
|
||||
| `version` | No | Schema version number (default: `1`) |
|
||||
| `schema` | Yes | Picoschema dict defining the fields |
|
||||
| `settings.validation` | No | Validation mode (default: `warn`) |
|
||||
|
||||
Schema notes are regular notes — they show up in search, can have observations and relations, and participate in the knowledge graph.
|
||||
|
||||
### Validation Modes
|
||||
|
||||
| Mode | Behavior |
|
||||
|------|----------|
|
||||
| `warn` | Warnings in output, doesn't block (default) |
|
||||
| `strict` | Errors that block sync, for CI/CD enforcement |
|
||||
| `off` | No validation |
|
||||
|
||||
### Validation Output
|
||||
|
||||
```
|
||||
$ bm schema validate people/ada-lovelace.md
|
||||
|
||||
⚠ Person schema validation:
|
||||
- Missing required field: name (expected [name] observation)
|
||||
- Missing optional field: role
|
||||
- Missing optional field: works_at (no relation found)
|
||||
|
||||
ℹ Unmatched observations: [fact] ×2, [born] ×1
|
||||
ℹ Unmatched relations: collaborated_with
|
||||
```
|
||||
|
||||
"Unmatched" items are informational — observations and relations the schema doesn't cover.
|
||||
|
||||
### Schema Inference
|
||||
|
||||
Generate schemas from existing notes by analyzing observation and relation frequency:
|
||||
|
||||
```
|
||||
$ bm schema infer Person
|
||||
|
||||
Analyzing 30 notes with type: Person...
|
||||
|
||||
Observations found:
|
||||
[name] 30/30 100% → name: string
|
||||
[role] 27/30 90% → role?: string
|
||||
[expertise] 18/30 60% → expertise?(array): string
|
||||
[email] 8/30 27% → email?: string
|
||||
|
||||
Relations found:
|
||||
works_at 22/30 73% → works_at?: Organization
|
||||
|
||||
Suggested schema:
|
||||
name: string, full name
|
||||
role?: string, job title
|
||||
expertise?(array): string, areas of knowledge
|
||||
email?: string, contact email
|
||||
works_at?: Organization, employer
|
||||
|
||||
Save to schema/Person.md? [y/n]
|
||||
```
|
||||
|
||||
Frequency thresholds:
|
||||
- **100% present** → required field
|
||||
- **25%+ present** → optional field
|
||||
- **Below 25%** → excluded from suggestion
|
||||
|
||||
### Schema Drift Detection
|
||||
|
||||
Track how usage patterns shift over time:
|
||||
|
||||
```
|
||||
$ bm schema diff Person
|
||||
|
||||
Schema drift detected:
|
||||
|
||||
+ expertise: now in 81% of notes (was 12%)
|
||||
- department: dropped to 3% of notes
|
||||
~ works_at: cardinality changed (one → many)
|
||||
|
||||
Update schema? [y/n/review]
|
||||
```
|
||||
|
||||
## Complete Examples
|
||||
|
||||
### Simple Note (No Schema)
|
||||
|
||||
```markdown
|
||||
---
|
||||
title: Project Ideas
|
||||
type: note
|
||||
tags: [ideas, brainstorm]
|
||||
---
|
||||
|
||||
# Project Ideas
|
||||
|
||||
## Observations
|
||||
- [idea] Build a CLI tool for markdown linting #tooling
|
||||
- [idea] Create a recipe knowledge base #cooking
|
||||
- [priority] Focus on developer tools first (Q1 goal)
|
||||
|
||||
## Relations
|
||||
- inspired_by [[Developer Workflow Research]]
|
||||
- part_of [[Q1 Planning]]
|
||||
```
|
||||
|
||||
### Schema-Validated Note
|
||||
|
||||
Schema at `schema/Person.md`:
|
||||
|
||||
```yaml
|
||||
---
|
||||
title: Person
|
||||
type: schema
|
||||
entity: Person
|
||||
version: 1
|
||||
schema:
|
||||
name: string, full name
|
||||
role?: string, job title or position
|
||||
works_at?: Organization, employer
|
||||
expertise?(array): string, areas of knowledge
|
||||
email?: string, contact email
|
||||
settings:
|
||||
validation: warn
|
||||
---
|
||||
|
||||
# Person
|
||||
|
||||
A human individual in the knowledge graph.
|
||||
```
|
||||
|
||||
Note at `people/paul-graham.md`:
|
||||
|
||||
```markdown
|
||||
---
|
||||
title: Paul Graham
|
||||
type: Person
|
||||
tags: [startups, essays, lisp]
|
||||
---
|
||||
|
||||
# Paul Graham
|
||||
|
||||
## Observations
|
||||
- [name] Paul Graham
|
||||
- [role] Essayist and investor
|
||||
- [expertise] Startups
|
||||
- [expertise] Lisp
|
||||
- [expertise] Essay writing
|
||||
- [fact] Created Viaweb, the first web app
|
||||
|
||||
## Relations
|
||||
- works_at [[Y Combinator]]
|
||||
- authored [[Hackers and Painters]]
|
||||
```
|
||||
|
||||
The `[fact]` observation and `authored` relation are not in the schema — they're valid, just unmatched. The schema only checks that `[name]` exists (required) and looks for optional fields like `[role]`, `[expertise]`, and `works_at`.
|
||||
|
||||
### Inline Schema Note
|
||||
|
||||
```markdown
|
||||
---
|
||||
title: Team Standup 2024-01-15
|
||||
type: meeting
|
||||
schema:
|
||||
attendees(array): string, who was there
|
||||
decisions(array): string, what was decided
|
||||
action_items(array): string, follow-ups
|
||||
blockers?(array): string, anything stuck
|
||||
---
|
||||
|
||||
# Team Standup 2024-01-15
|
||||
|
||||
## Observations
|
||||
- [attendees] Paul
|
||||
- [attendees] Sarah
|
||||
- [decisions] Ship v2 by Friday
|
||||
- [action_items] Paul to review PR #42
|
||||
- [blockers] Waiting on API credentials
|
||||
```
|
||||
+2
-67
@@ -8,71 +8,6 @@
|
||||
|
||||
## Reporting a Vulnerability
|
||||
|
||||
If you find a vulnerability, please contact hello@basicmachines.co.
|
||||
Use this section to tell people how to report a vulnerability.
|
||||
|
||||
Please do not open a public GitHub issue for security vulnerabilities. We aim
|
||||
to respond within 72 hours and will coordinate a fix and disclosure timeline
|
||||
with you.
|
||||
|
||||
## Threat Model
|
||||
|
||||
Basic Memory is a local-first MCP server that reads and writes markdown files
|
||||
inside configured project directories. It runs on your machine with your user
|
||||
permissions, so local configuration deserves the same care as any other
|
||||
developer tool that can access your files.
|
||||
|
||||
### What Basic Memory Controls
|
||||
|
||||
- Filesystem-touching tools validate paths against the configured project root
|
||||
with `validate_project_path()`, resolved paths, and `Path.is_relative_to()`.
|
||||
Path traversal attempts such as `../../etc/passwd` are blocked at this layer.
|
||||
- Scan optimizations in `sync_service.py` call `find` through
|
||||
`asyncio.create_subprocess_exec()` with explicit argument lists. Project paths
|
||||
are passed as data, not interpolated into shell strings.
|
||||
- Auto-update code uses hardcoded commands, list-form arguments, and
|
||||
`stdin=DEVNULL`. User-controlled strings do not reach a shell there.
|
||||
|
||||
### MCP Client-Side Risk
|
||||
|
||||
Recent MCP ecosystem research has highlighted a client-side pattern where an
|
||||
MCP host can be configured to run arbitrary commands as "servers." That risk is
|
||||
in the host configuration, not in notes or Basic Memory tool input.
|
||||
|
||||
The recommended Basic Memory MCP configuration uses a known command with
|
||||
explicit arguments:
|
||||
|
||||
```json
|
||||
{
|
||||
"mcpServers": {
|
||||
"basic-memory": {
|
||||
"command": "uvx",
|
||||
"args": ["basic-memory", "mcp"]
|
||||
}
|
||||
}
|
||||
}
|
||||
```
|
||||
|
||||
Only add MCP server entries from sources you trust. Avoid inline shell scripts
|
||||
or command strings copied from untrusted sources. Treat third-party MCP server
|
||||
configuration with the same scrutiny as any locally executed program.
|
||||
|
||||
Related ecosystem context:
|
||||
|
||||
- OX Security: The Mother of All AI Supply Chains
|
||||
- CSO Online: RCE by design: MCP architectural choice haunts AI agent ecosystem
|
||||
|
||||
### Out Of Scope
|
||||
|
||||
- Basic Memory does not execute note content as code. Notes are returned as
|
||||
data to the LLM.
|
||||
- Basic Memory does not open network ports by default. The MCP server uses
|
||||
stdio; the optional REST API is intended for localhost use.
|
||||
- Basic Memory is designed for single-user local knowledge bases and does not
|
||||
implement access controls between operating-system users.
|
||||
|
||||
## Secure Configuration Checklist
|
||||
|
||||
- MCP config `command` points to `uvx` or a trusted binary, not a shell string.
|
||||
- Project paths in Basic Memory config come from trusted local configuration.
|
||||
- If exposing the REST API, bind it only to localhost.
|
||||
- Review any third-party MCP servers before adding them to your host config.
|
||||
If you find a vulnerability, please contact hello@basicmachines.co
|
||||
|
||||
@@ -1,8 +1,5 @@
|
||||
# Docker Compose configuration for Basic Memory with PostgreSQL
|
||||
# Use this for local development and testing with Postgres backend.
|
||||
#
|
||||
# The Postgres backend requires the pgvector extension (semantic search).
|
||||
# This image bundles pgvector; plain postgres:17 will not work for vector search.
|
||||
# Use this for local development and testing with Postgres backend
|
||||
#
|
||||
# Usage:
|
||||
# docker-compose -f docker-compose-postgres.yml up -d
|
||||
@@ -10,7 +7,7 @@
|
||||
|
||||
services:
|
||||
postgres:
|
||||
image: pgvector/pgvector:pg17
|
||||
image: postgres:17
|
||||
container_name: basic-memory-postgres
|
||||
environment:
|
||||
# Local development/test credentials - NOT for production
|
||||
|
||||
+1
-3
@@ -17,9 +17,7 @@ services:
|
||||
volumes:
|
||||
|
||||
# Persistent storage for configuration and database
|
||||
# Container runs as `appuser` (Dockerfile USER directive), so the CLI
|
||||
# config dir lives under /home/appuser, not /root.
|
||||
- basic-memory-config:/home/appuser/.basic-memory:rw
|
||||
- basic-memory-config:/root/.basic-memory:rw
|
||||
|
||||
# Mount your knowledge directory (required)
|
||||
# Change './knowledge' to your actual Obsidian vault or knowledge directory
|
||||
|
||||
@@ -1,442 +0,0 @@
|
||||
# Basic Memory Architecture
|
||||
|
||||
This document describes the architectural patterns and composition structure of Basic Memory.
|
||||
|
||||
## Overview
|
||||
|
||||
Basic Memory is a local-first knowledge management system with three entrypoints:
|
||||
- **API** - FastAPI REST server for HTTP access
|
||||
- **MCP** - Model Context Protocol server for LLM integration
|
||||
- **CLI** - Typer command-line interface
|
||||
|
||||
Each entrypoint uses a **composition root** pattern to manage configuration and dependencies.
|
||||
|
||||
## Composition Roots
|
||||
|
||||
### What is a Composition Root?
|
||||
|
||||
A composition root is the single place in an application where dependencies are wired together. In Basic Memory, each entrypoint has its own composition root that:
|
||||
|
||||
1. Reads configuration from `ConfigManager`
|
||||
2. Resolves runtime mode (local/test)
|
||||
3. Creates and provides dependencies to downstream code
|
||||
|
||||
**Key principle**: Only composition roots read global configuration. All other modules receive configuration explicitly.
|
||||
|
||||
### Container Structure
|
||||
|
||||
Each entrypoint has a container dataclass in its package:
|
||||
|
||||
```
|
||||
src/basic_memory/
|
||||
├── api/
|
||||
│ └── container.py # ApiContainer
|
||||
├── mcp/
|
||||
│ └── container.py # McpContainer
|
||||
├── cli/
|
||||
│ └── container.py # CliContainer
|
||||
└── runtime.py # RuntimeMode enum and resolver
|
||||
```
|
||||
|
||||
### Container Pattern
|
||||
|
||||
All containers follow the same structure:
|
||||
|
||||
```python
|
||||
@dataclass
|
||||
class Container:
|
||||
config: BasicMemoryConfig
|
||||
mode: RuntimeMode
|
||||
|
||||
@classmethod
|
||||
def create(cls) -> "Container":
|
||||
"""Create container by reading ConfigManager."""
|
||||
config = ConfigManager().config
|
||||
mode = resolve_runtime_mode(is_test_env=config.is_test_env)
|
||||
return cls(config=config, mode=mode)
|
||||
|
||||
@property
|
||||
def some_computed_property(self) -> bool:
|
||||
"""Derived values based on config and mode."""
|
||||
return self.mode.is_local and self.config.some_setting
|
||||
|
||||
# Module-level singleton
|
||||
_container: Container | None = None
|
||||
|
||||
def get_container() -> Container:
|
||||
if _container is None:
|
||||
raise RuntimeError("Container not initialized")
|
||||
return _container
|
||||
|
||||
def set_container(container: Container) -> None:
|
||||
global _container
|
||||
_container = container
|
||||
```
|
||||
|
||||
### Runtime Mode Resolution
|
||||
|
||||
The `RuntimeMode` enum centralizes mode detection:
|
||||
|
||||
```python
|
||||
class RuntimeMode(Enum):
|
||||
LOCAL = "local"
|
||||
CLOUD = "cloud"
|
||||
TEST = "test"
|
||||
|
||||
@property
|
||||
def is_cloud(self) -> bool:
|
||||
return self == RuntimeMode.CLOUD
|
||||
|
||||
@property
|
||||
def is_local(self) -> bool:
|
||||
return self == RuntimeMode.LOCAL
|
||||
|
||||
@property
|
||||
def is_test(self) -> bool:
|
||||
return self == RuntimeMode.TEST
|
||||
```
|
||||
|
||||
Resolution follows this precedence in local app flows: **TEST > LOCAL**
|
||||
|
||||
```python
|
||||
def resolve_runtime_mode(is_test_env: bool) -> RuntimeMode:
|
||||
if is_test_env:
|
||||
return RuntimeMode.TEST
|
||||
return RuntimeMode.LOCAL
|
||||
```
|
||||
|
||||
**Note**: `RuntimeMode` determines global behavior (e.g., whether to start file sync).
|
||||
Per-project routing is orthogonal: individual projects can be set to `cloud` mode via `ProjectMode`,
|
||||
which affects client routing in `get_client(project_name=...)` without changing global runtime mode.
|
||||
`RuntimeMode.CLOUD` may remain for compatibility, but standard local runtime resolution does not select it.
|
||||
|
||||
## Dependencies Package
|
||||
|
||||
### Structure
|
||||
|
||||
The `deps/` package provides FastAPI dependencies organized by feature:
|
||||
|
||||
```
|
||||
src/basic_memory/deps/
|
||||
├── __init__.py # Re-exports for backwards compatibility
|
||||
├── config.py # Configuration access
|
||||
├── db.py # Database/session management
|
||||
├── projects.py # Project resolution
|
||||
├── repositories.py # Data access layer
|
||||
├── services.py # Business logic layer
|
||||
└── importers.py # Import functionality
|
||||
```
|
||||
|
||||
### Usage in Routers
|
||||
|
||||
```python
|
||||
from basic_memory.deps.services import get_entity_service
|
||||
from basic_memory.deps.projects import get_project_config
|
||||
|
||||
@router.get("/entities/{id}")
|
||||
async def get_entity(
|
||||
id: int,
|
||||
entity_service: EntityService = Depends(get_entity_service),
|
||||
project: ProjectConfig = Depends(get_project_config),
|
||||
):
|
||||
return await entity_service.get(id)
|
||||
```
|
||||
|
||||
### Backwards Compatibility
|
||||
|
||||
The old `deps.py` file still exists as a thin re-export shim:
|
||||
|
||||
```python
|
||||
# deps.py - backwards compatibility shim
|
||||
from basic_memory.deps import *
|
||||
```
|
||||
|
||||
New code should import from specific submodules (`basic_memory.deps.services`) for clarity.
|
||||
|
||||
## MCP Tools Architecture
|
||||
|
||||
### Typed API Clients
|
||||
|
||||
MCP tools communicate with the API through typed clients that encapsulate HTTP paths and response validation:
|
||||
|
||||
```
|
||||
src/basic_memory/mcp/clients/
|
||||
├── __init__.py # Re-exports all clients
|
||||
├── base.py # BaseClient with common logic
|
||||
├── knowledge.py # KnowledgeClient - entity CRUD
|
||||
├── search.py # SearchClient - search operations
|
||||
├── memory.py # MemoryClient - context building
|
||||
├── directory.py # DirectoryClient - directory listing
|
||||
├── resource.py # ResourceClient - resource reading
|
||||
└── project.py # ProjectClient - project management
|
||||
```
|
||||
|
||||
### Client Pattern
|
||||
|
||||
Each client encapsulates API paths and validates responses:
|
||||
|
||||
```python
|
||||
class KnowledgeClient(BaseClient):
|
||||
"""Client for knowledge/entity operations."""
|
||||
|
||||
async def resolve_entity(self, identifier: str) -> int:
|
||||
"""Resolve identifier to entity ID."""
|
||||
response = await call_get(
|
||||
self.http_client,
|
||||
f"{self._base_path}/resolve/{identifier}",
|
||||
)
|
||||
return int(response.text)
|
||||
|
||||
async def get_entity(self, entity_id: int) -> EntityResponse:
|
||||
"""Get entity by ID."""
|
||||
response = await call_get(
|
||||
self.http_client,
|
||||
f"{self._base_path}/entities/{entity_id}",
|
||||
)
|
||||
return EntityResponse.model_validate(response.json())
|
||||
```
|
||||
|
||||
### Tool → Client → API Flow
|
||||
|
||||
```
|
||||
MCP Tool (thin adapter)
|
||||
↓
|
||||
Typed Client (encapsulates paths, validates responses)
|
||||
↓
|
||||
HTTP API (FastAPI router)
|
||||
↓
|
||||
Service Layer (business logic)
|
||||
↓
|
||||
Repository Layer (data access)
|
||||
```
|
||||
|
||||
Example tool using typed client:
|
||||
|
||||
```python
|
||||
@mcp.tool()
|
||||
async def search_notes(
|
||||
query: str,
|
||||
project: str | None = None,
|
||||
metadata_filters: dict | None = None,
|
||||
tags: list[str] | None = None,
|
||||
status: str | None = None,
|
||||
) -> SearchResponse:
|
||||
async with get_project_client(project, context) as (client, active_project):
|
||||
# Import client inside function to avoid circular imports
|
||||
from basic_memory.mcp.clients import SearchClient
|
||||
from basic_memory.schemas.search import SearchQuery
|
||||
|
||||
search_query = SearchQuery(
|
||||
text=query,
|
||||
metadata_filters=metadata_filters,
|
||||
tags=tags,
|
||||
status=status,
|
||||
)
|
||||
search_client = SearchClient(client, active_project.external_id)
|
||||
return await search_client.search(search_query.model_dump())
|
||||
```
|
||||
|
||||
### Per-Project Client Routing
|
||||
|
||||
`get_project_client()` from `mcp/project_context.py` is an async context manager that:
|
||||
1. Resolves the project name from config (no network call)
|
||||
2. Creates the correctly-routed client based on the project's mode (local ASGI or cloud HTTP with API key)
|
||||
3. Validates the project via the API
|
||||
4. Yields `(client, active_project)` tuple
|
||||
|
||||
This solves the bootstrap problem: you need the project name to choose the right client (local vs cloud), but you need the client to validate the project exists.
|
||||
|
||||
```python
|
||||
from basic_memory.mcp.project_context import get_project_client
|
||||
|
||||
async with get_project_client(project, context) as (client, active_project):
|
||||
# client is routed based on project's mode (local or cloud)
|
||||
# active_project is validated via the API
|
||||
...
|
||||
```
|
||||
|
||||
## Sync Coordination
|
||||
|
||||
### SyncCoordinator
|
||||
|
||||
The `SyncCoordinator` centralizes sync/watch lifecycle management:
|
||||
|
||||
```python
|
||||
@dataclass
|
||||
class SyncCoordinator:
|
||||
"""Coordinates file sync and watch operations."""
|
||||
|
||||
status: SyncStatus = SyncStatus.NOT_STARTED
|
||||
sync_task: asyncio.Task | None = None
|
||||
watch_service: WatchService | None = None
|
||||
|
||||
async def start(self, ...):
|
||||
"""Start sync and watch operations."""
|
||||
|
||||
async def stop(self):
|
||||
"""Stop all sync operations gracefully."""
|
||||
|
||||
def get_status_info(self) -> dict:
|
||||
"""Get current sync status for observability."""
|
||||
```
|
||||
|
||||
### Status Enum
|
||||
|
||||
```python
|
||||
class SyncStatus(Enum):
|
||||
NOT_STARTED = "not_started"
|
||||
STARTING = "starting"
|
||||
RUNNING = "running"
|
||||
STOPPING = "stopping"
|
||||
STOPPED = "stopped"
|
||||
ERROR = "error"
|
||||
```
|
||||
|
||||
## Project Resolution
|
||||
|
||||
### ProjectResolver
|
||||
|
||||
Unified project selection across all entrypoints:
|
||||
|
||||
```python
|
||||
class ProjectResolver:
|
||||
"""Resolves which project to use based on context."""
|
||||
|
||||
def resolve(
|
||||
self,
|
||||
explicit_project: str | None = None,
|
||||
) -> ResolvedProject:
|
||||
"""Resolve project using three-tier hierarchy:
|
||||
1. Explicit project parameter
|
||||
2. Default project from config
|
||||
3. Single available project
|
||||
"""
|
||||
```
|
||||
|
||||
### Resolution Modes
|
||||
|
||||
```python
|
||||
class ResolutionMode(Enum):
|
||||
EXPLICIT = "explicit" # User specified project
|
||||
DEFAULT = "default" # Using configured default
|
||||
SINGLE_PROJECT = "single" # Only one project exists
|
||||
FALLBACK = "fallback" # Using first available
|
||||
```
|
||||
|
||||
## Testing Patterns
|
||||
|
||||
### Container Testing
|
||||
|
||||
Each container has corresponding tests:
|
||||
|
||||
```
|
||||
tests/
|
||||
├── api/test_api_container.py
|
||||
├── mcp/test_mcp_container.py
|
||||
└── cli/test_cli_container.py
|
||||
```
|
||||
|
||||
Tests verify:
|
||||
- Container creation from config
|
||||
- Runtime mode properties
|
||||
- Container accessor functions (get/set)
|
||||
|
||||
### Mocking Typed Clients
|
||||
|
||||
When testing MCP tools, mock at the client level:
|
||||
|
||||
```python
|
||||
def test_search_notes(monkeypatch):
|
||||
import basic_memory.mcp.clients as clients_mod
|
||||
|
||||
class MockSearchClient:
|
||||
async def search(self, query):
|
||||
return SearchResponse(results=[...])
|
||||
|
||||
monkeypatch.setattr(clients_mod, "SearchClient", MockSearchClient)
|
||||
```
|
||||
|
||||
## Design Principles
|
||||
|
||||
### 1. Explicit Dependencies
|
||||
|
||||
Modules receive configuration explicitly rather than reading globals:
|
||||
|
||||
```python
|
||||
# Good - explicit injection
|
||||
async def sync_files(config: BasicMemoryConfig):
|
||||
...
|
||||
|
||||
# Avoid - hidden global access
|
||||
async def sync_files():
|
||||
config = ConfigManager().config # Hidden coupling
|
||||
```
|
||||
|
||||
### 2. Single Responsibility
|
||||
|
||||
Each layer has a clear responsibility:
|
||||
- **Containers**: Wire dependencies
|
||||
- **Clients**: Encapsulate HTTP communication
|
||||
- **Services**: Business logic
|
||||
- **Repositories**: Data access
|
||||
- **Tools/Routers**: Thin adapters
|
||||
|
||||
### 3. Deferred Imports
|
||||
|
||||
To avoid circular imports, typed clients are imported inside functions:
|
||||
|
||||
```python
|
||||
async def my_tool():
|
||||
async with get_client() as client:
|
||||
# Import here to avoid circular dependency
|
||||
from basic_memory.mcp.clients import KnowledgeClient
|
||||
|
||||
knowledge_client = KnowledgeClient(client, project_id)
|
||||
```
|
||||
|
||||
### 4. Backwards Compatibility
|
||||
|
||||
When refactoring, maintain backwards compatibility via shims:
|
||||
|
||||
```python
|
||||
# Old module becomes a shim
|
||||
from basic_memory.new_location import *
|
||||
|
||||
# Docstring explains migration path
|
||||
"""
|
||||
DEPRECATED: Import from basic_memory.new_location instead.
|
||||
This shim will be removed in a future version.
|
||||
"""
|
||||
```
|
||||
|
||||
## File Organization
|
||||
|
||||
```
|
||||
src/basic_memory/
|
||||
├── api/
|
||||
│ ├── container.py # API composition root
|
||||
│ ├── routers/ # FastAPI routers
|
||||
│ └── ...
|
||||
├── mcp/
|
||||
│ ├── container.py # MCP composition root
|
||||
│ ├── clients/ # Typed API clients
|
||||
│ ├── tools/ # MCP tool definitions
|
||||
│ └── server.py # MCP server setup
|
||||
├── cli/
|
||||
│ ├── container.py # CLI composition root
|
||||
│ ├── app.py # Typer app
|
||||
│ └── commands/ # CLI command groups
|
||||
├── deps/
|
||||
│ ├── config.py # Config dependencies
|
||||
│ ├── db.py # Database dependencies
|
||||
│ ├── projects.py # Project dependencies
|
||||
│ ├── repositories.py # Repository dependencies
|
||||
│ ├── services.py # Service dependencies
|
||||
│ └── importers.py # Importer dependencies
|
||||
├── sync/
|
||||
│ ├── coordinator.py # SyncCoordinator
|
||||
│ └── ...
|
||||
├── runtime.py # RuntimeMode resolution
|
||||
├── project_resolver.py # Unified project selection
|
||||
└── config.py # Configuration management
|
||||
```
|
||||
+3
-3
@@ -111,7 +111,7 @@ You can run Basic Memory CLI commands inside the container using `docker exec`:
|
||||
docker exec basic-memory-server basic-memory status
|
||||
|
||||
# Sync files
|
||||
docker exec basic-memory-server basic-memory reindex
|
||||
docker exec basic-memory-server basic-memory sync
|
||||
|
||||
# Show help
|
||||
docker exec basic-memory-server basic-memory --help
|
||||
@@ -137,7 +137,7 @@ When using Docker volumes, you'll need to configure projects to point to your mo
|
||||
|
||||
3. **Sync the new project:**
|
||||
```bash
|
||||
docker exec basic-memory-server basic-memory reindex
|
||||
docker exec basic-memory-server basic-memory sync
|
||||
```
|
||||
|
||||
### Example: Setting up an Obsidian Vault
|
||||
@@ -157,7 +157,7 @@ docker exec basic-memory-server basic-memory project create obsidian /app/data
|
||||
docker exec basic-memory-server basic-memory project set-default obsidian
|
||||
|
||||
# Sync to index all files
|
||||
docker exec basic-memory-server basic-memory reindex
|
||||
docker exec basic-memory-server basic-memory sync
|
||||
```
|
||||
|
||||
### Environment Variables
|
||||
|
||||
@@ -1,64 +0,0 @@
|
||||
# Basic Memory Engineering Style
|
||||
|
||||
Style is how we make code easier to verify. Prefer explicit, typed, local-first code that
|
||||
preserves the file system as the source of truth while keeping the database, API, and MCP
|
||||
surfaces in sync.
|
||||
|
||||
## Design Center
|
||||
|
||||
- Basic Memory is local-first. Markdown files are the durable source; SQLite/Postgres indexes
|
||||
are derived state that should be rebuilt or reconciled from files when needed.
|
||||
- Keep the existing boundary order: CLI/MCP/API entrypoints compose dependencies, services own
|
||||
business behavior, repositories own database access, and file services own filesystem writes.
|
||||
- MCP tools should remain atomic and composable. They should call API routers through typed MCP
|
||||
clients, not reach around into services.
|
||||
- Prefer small, explicit abstractions that match a real domain boundary. Avoid object
|
||||
hierarchies when a function, dataclass, type alias, or protocol describes the concept better.
|
||||
|
||||
## Types And Data
|
||||
|
||||
- Use full type annotations and Python 3.12 syntax. Introduce `type` aliases for repeated
|
||||
structured shapes, callback signatures, or domain concepts that would otherwise become
|
||||
anonymous `dict[str, Any]` values.
|
||||
- Use dataclasses for internal values, operation inputs, and service results. Prefer
|
||||
`frozen=True` when the value should not change and `slots=True` when identity/dynamic
|
||||
attributes are not needed.
|
||||
- Use Pydantic v2 at boundaries that validate, serialize, or deserialize data: API payloads,
|
||||
CLI/MCP schemas, configuration, and persistence-adjacent schemas.
|
||||
- Use narrow `Protocol`s when a caller needs a capability rather than a concrete repository or
|
||||
service. Keep protocols small enough that fake implementations in tests are obvious.
|
||||
- Avoid speculative `getattr`, broad casts, or `Any` as a way to paper over uncertainty. Read
|
||||
the model or schema definition and make the type relationship explicit.
|
||||
|
||||
## Control Flow And Resources
|
||||
|
||||
- Fail fast when an invariant is broken. Do not swallow exceptions, add warning-only error
|
||||
handling, or introduce fallback behavior unless the user explicitly agrees to that behavior.
|
||||
- Keep control flow simple and close to the domain decision. Push `if` statements up into the
|
||||
function that owns orchestration; keep leaf helpers focused on computation or one side effect.
|
||||
- Make async/resource boundaries visible with context managers and explicit lifecycles. Do not
|
||||
start background work without a clear owner, cancellation story, and verification path.
|
||||
- Keep file mutations centralized through the existing file utilities/services so checksum,
|
||||
atomic write, and index synchronization behavior stays coherent.
|
||||
|
||||
## Testing And Verification
|
||||
|
||||
- Use evidence-first testing, not mechanical TDD. For bugs and risky behavior, add or update a
|
||||
regression test that would catch the failure. For small documentation-only edits, use the
|
||||
relevant doc/repo hygiene checks.
|
||||
- Prefer tests that exercise real code paths. Use mocks, doubles, or `monkeypatch` only when
|
||||
the external boundary would be slow, nondeterministic, or impossible to trigger directly.
|
||||
- Keep coverage at 100% for new code. Use `# pragma: no cover` only for code that would require
|
||||
disproportionate mocking and is covered through an integration or runtime path.
|
||||
- Start with targeted commands, then widen as risk grows: focused pytest, `just fast-check`,
|
||||
`just doctor`, package checks for agent packaging changes, and full SQLite/Postgres gates
|
||||
when behavior crosses shared boundaries.
|
||||
|
||||
## Comments And Names
|
||||
|
||||
- Name values after the domain concept they carry: project, entity, permalink, tenant, route,
|
||||
checksum, observation, relation, batch, or index state.
|
||||
- Comments should say why a branch, invariant, retry, lifecycle, or compatibility constraint
|
||||
exists. Section headers are useful when a function or file has clear phases.
|
||||
- Avoid comments that restate the code. If a comment cannot explain a decision, simplify the
|
||||
code or improve the name instead.
|
||||
@@ -1,512 +0,0 @@
|
||||
# Note Format Reference
|
||||
|
||||
Every document in Basic Memory is a plain Markdown file. Files are the source of truth — changes to files automatically update the knowledge graph in the database. You maintain complete ownership, files work with git, and knowledge persists independently of any AI conversation.
|
||||
|
||||
## Document Structure
|
||||
|
||||
A note has three parts: YAML frontmatter, content (observations), and relations.
|
||||
|
||||
```markdown
|
||||
---
|
||||
title: Coffee Brewing Methods
|
||||
type: note
|
||||
tags: [coffee, brewing]
|
||||
permalink: coffee-brewing-methods
|
||||
---
|
||||
|
||||
# Coffee Brewing Methods
|
||||
|
||||
## Observations
|
||||
- [method] Pour over provides more flavor clarity than French press
|
||||
- [technique] Water temperature at 205°F extracts optimal compounds #brewing
|
||||
- [preference] Ethiopian beans work well with lighter roasts (personal experience)
|
||||
|
||||
## Relations
|
||||
- relates_to [[Coffee Bean Origins]]
|
||||
- requires [[Proper Grinding Technique]]
|
||||
- contrasts_with [[Tea Brewing Methods]]
|
||||
```
|
||||
|
||||
The `## Observations` and `## Relations` headings are conventional but not required — the parser detects observations and relations by their syntax patterns anywhere in the document.
|
||||
|
||||
## Frontmatter
|
||||
|
||||
YAML metadata between `---` fences at the top of the file.
|
||||
|
||||
| Field | Required | Default | Description |
|
||||
|-------|----------|---------|-------------|
|
||||
| `title` | No | filename stem | Used for linking and references. Auto-set from filename if missing. |
|
||||
| `type` | No | `note` | Entity type. Used for schema resolution and filtering. |
|
||||
| `tags` | No | `[]` | List or comma-separated string. Used for organization and search. |
|
||||
| `permalink` | No | generated from title | Stable identifier. Persists even if the file moves. |
|
||||
| `schema` | No | none | Schema attachment — dict (inline), string (reference), or omitted (implicit). |
|
||||
|
||||
Custom fields are allowed. Any key not in the standard set is stored as `entity_metadata` and indexed for search and filtering.
|
||||
|
||||
```yaml
|
||||
---
|
||||
title: Paul Graham
|
||||
type: Person
|
||||
tags: [startups, essays, lisp]
|
||||
permalink: paul-graham
|
||||
status: active
|
||||
source: wikipedia
|
||||
---
|
||||
```
|
||||
|
||||
Here `status` and `source` are custom fields stored in `entity_metadata`.
|
||||
|
||||
### Frontmatter Value Handling
|
||||
|
||||
YAML automatically converts some values to native types. Basic Memory normalizes them:
|
||||
|
||||
- Date strings (`2025-10-24`) → kept as ISO format strings
|
||||
- Numbers (`1.0`) → converted to strings
|
||||
- Booleans (`true`) → converted to strings (`"True"`)
|
||||
- Lists and dicts → preserved, items normalized recursively
|
||||
|
||||
This prevents errors when downstream code expects string values.
|
||||
|
||||
## Observations
|
||||
|
||||
An observation is a categorized fact about the entity. Written as a Markdown list item.
|
||||
|
||||
**Syntax:**
|
||||
|
||||
```
|
||||
- [category] content text #tag1 #tag2 (context)
|
||||
```
|
||||
|
||||
| Part | Required | Description |
|
||||
|------|----------|-------------|
|
||||
| `[category]` | Yes | Classification in square brackets. Any text except `[]()` chars. |
|
||||
| content | Yes | The fact or statement. |
|
||||
| `#tags` | No | Inline tags. Space-separated, each starting with `#`. |
|
||||
| `(context)` | No | Parenthesized text at end of line. Supporting details or source. |
|
||||
|
||||
### Examples
|
||||
|
||||
```markdown
|
||||
- [tech] Uses SQLite for storage #database
|
||||
- [design] Follows local-first architecture #architecture
|
||||
- [decision] Selected bcrypt for passwords #security (based on OWASP audit)
|
||||
- [name] Paul Graham
|
||||
- [expertise] Startups
|
||||
- [expertise] Lisp
|
||||
- [expertise] Essay writing
|
||||
```
|
||||
|
||||
Array-like fields use repeated categories — multiple `[expertise]` observations above.
|
||||
|
||||
### What Is Not an Observation
|
||||
|
||||
The parser excludes these list item patterns:
|
||||
|
||||
| Pattern | Example | Reason |
|
||||
|---------|---------|--------|
|
||||
| Checkboxes | `- [ ] Todo item`, `- [x] Done`, `- [-] Cancelled` | Task list syntax |
|
||||
| Markdown links | `- [text](url)` | URL link syntax |
|
||||
| Bare wiki links | `- [[Target]]` | Treated as a `links_to` relation instead |
|
||||
|
||||
A list item with `#tags` but no `[category]` is still parsed — the tags are extracted and the category defaults to `Note`.
|
||||
|
||||
## Relations
|
||||
|
||||
Relations connect documents to form the knowledge graph. There are two kinds.
|
||||
|
||||
### Explicit Relations
|
||||
|
||||
Written as list items with a relation type and a `[[wiki link]]` target. Unquoted
|
||||
relation types are single tokens. Quote relation types that contain spaces.
|
||||
|
||||
**Syntax:**
|
||||
|
||||
```
|
||||
- relation_type [[Target Entity]] (context)
|
||||
- "multi word relation type" [[Target Entity]] (context)
|
||||
- 'multi word relation type' [[Target Entity]] (context)
|
||||
```
|
||||
|
||||
| Part | Required | Description |
|
||||
|------|----------|-------------|
|
||||
| `relation_type` | Yes | Single unquoted token before `[[`, or quoted text for multi-word labels. |
|
||||
| `[[Target]]` | Yes | Wiki link to the target entity. Matched by title or permalink. |
|
||||
| `(context)` | No | Parenthesized text after `]]`. Supporting details. |
|
||||
|
||||
### Examples
|
||||
|
||||
Explicit relations:
|
||||
|
||||
```markdown
|
||||
- implements [[Search Design]]
|
||||
- depends_on [[Database Schema]]
|
||||
- works_at [[Y Combinator]] (co-founder)
|
||||
- "based on" [[Customer Interview]]
|
||||
- 'in response to' [[Incident Review]]
|
||||
```
|
||||
|
||||
Bare wiki links and prose list items create implicit `links_to` relations:
|
||||
|
||||
```markdown
|
||||
- [[Some Entity]]
|
||||
- some other thing [[Some Entity]]
|
||||
```
|
||||
|
||||
Both examples above create `links_to [[Some Entity]]`. Use quotes when the words before
|
||||
`[[` are meant to be a multi-word relation type.
|
||||
|
||||
Common relation types:
|
||||
- `implements`, `depends_on`, `relates_to`, `inspired_by`
|
||||
- `extends`, `part_of`, `contains`, `pairs_with`
|
||||
- `works_at`, `authored`, `collaborated_with`
|
||||
|
||||
Any single-token text or quoted text works as a relation type. These are conventions,
|
||||
not a fixed set.
|
||||
|
||||
### Inline References
|
||||
|
||||
Wiki links appearing in regular prose create implicit `links_to` relations. This includes
|
||||
list items that do not match the explicit relation grammar above.
|
||||
|
||||
```markdown
|
||||
This builds on [[Core Design]] and uses [[Utility Functions]].
|
||||
- We should revisit [[Search Design]] after the API changes.
|
||||
```
|
||||
|
||||
This creates three relations: `links_to [[Core Design]]`, `links_to [[Utility Functions]]`,
|
||||
and `links_to [[Search Design]]`.
|
||||
|
||||
### Forward References
|
||||
|
||||
Relations can link to entities that don't exist yet. Basic Memory resolves them when the target is created.
|
||||
|
||||
## Permalinks and memory:// URLs
|
||||
|
||||
Every document has a unique **permalink** — a stable identifier derived from its title. You can set one explicitly in frontmatter, or let the system generate it.
|
||||
|
||||
```yaml
|
||||
permalink: auth-approaches-2024
|
||||
```
|
||||
|
||||
Permalinks form the basis of `memory://` URLs:
|
||||
|
||||
```
|
||||
memory://auth-approaches-2024 # By permalink
|
||||
memory://Authentication Approaches # By title (auto-resolves)
|
||||
memory://project/auth-approaches # By path
|
||||
```
|
||||
|
||||
Pattern matching is supported:
|
||||
|
||||
```
|
||||
memory://auth* # Starts with "auth"
|
||||
memory://*/approaches # Ends with "approaches"
|
||||
memory://project/*/requirements # Nested wildcard
|
||||
```
|
||||
|
||||
## Schemas
|
||||
|
||||
Schemas declare the expected structure of a note — which observation categories and relation types a well-formed note should have. They use Picoschema, a compact notation from Google's Dotprompt that fits naturally in YAML frontmatter.
|
||||
|
||||
### Picoschema Syntax
|
||||
|
||||
```yaml
|
||||
schema:
|
||||
name: string, full name # required field with description
|
||||
email?: string, contact email # ? = optional
|
||||
role?: string, job title
|
||||
works_at?: Organization, employer # capitalized type = entity reference
|
||||
tags?(array): string, categories # array of type
|
||||
status?(enum): [active, inactive] # enum with allowed values
|
||||
metadata?(object): # nested object
|
||||
updated_at?: string
|
||||
source?: string
|
||||
```
|
||||
|
||||
| Notation | Meaning | Example |
|
||||
|----------|---------|---------|
|
||||
| `field: type` | Required field | `name: string` |
|
||||
| `field?: type` | Optional field | `role?: string` |
|
||||
| `field(array): type` | Array of values | `expertise(array): string` |
|
||||
| `field?(enum): [vals]` | Enum with allowed values | `status?(enum): [active, inactive]` |
|
||||
| `field?(object):` | Nested object with sub-fields | `metadata?(object):` |
|
||||
| `, description` | Description after comma | `name: string, full name` |
|
||||
| `EntityName` | Capitalized type = entity reference | `works_at?: Organization` |
|
||||
|
||||
**Scalar types:** `string`, `integer`, `number`, `boolean`, `any`
|
||||
|
||||
Any type not in that set whose first letter is uppercase is treated as an entity reference (a relation target).
|
||||
|
||||
### Schema-to-Note Mapping
|
||||
|
||||
Schemas validate against existing observation/relation syntax. Note authors don't learn new syntax.
|
||||
|
||||
| Schema Declaration | Maps To | Example in Note |
|
||||
|--------------------|---------|-----------------|
|
||||
| `field: string` | Observation `[field] value` | `- [name] Paul Graham` |
|
||||
| `field?(array): string` | Multiple `[field]` observations | `- [expertise] Lisp` (repeated) |
|
||||
| `field?: EntityType` | Relation `field [[Target]]` | `- works_at [[Y Combinator]]` |
|
||||
| `field?(array): EntityType` | Multiple `field` relations | `- authored [[Book]]` (repeated) |
|
||||
| `tags` | Frontmatter `tags` array | `tags: [startups, essays]` |
|
||||
| `field?(enum): [vals]` | Observation `[field] value` where value is in the set | `- [status] active` |
|
||||
|
||||
Observations and relations not covered by the schema are valid — schemas describe a subset, not a straitjacket.
|
||||
|
||||
### Schema Attachment
|
||||
|
||||
Three ways to attach a schema to a note, resolved in priority order:
|
||||
|
||||
**1. Inline schema** — `schema` is a dict in frontmatter:
|
||||
|
||||
```yaml
|
||||
---
|
||||
title: Team Standup 2024-01-15
|
||||
type: meeting
|
||||
schema:
|
||||
attendees(array): string, who was there
|
||||
decisions(array): string, what was decided
|
||||
action_items(array): string, follow-ups
|
||||
blockers?(array): string, anything stuck
|
||||
---
|
||||
```
|
||||
|
||||
Good for one-off structured notes or prototyping a schema before extracting it.
|
||||
|
||||
**2. Explicit reference** — `schema` is a string naming a schema note:
|
||||
|
||||
```yaml
|
||||
---
|
||||
title: Basic Memory
|
||||
schema: SoftwareProject
|
||||
---
|
||||
```
|
||||
|
||||
or by permalink:
|
||||
|
||||
```yaml
|
||||
---
|
||||
title: LLM Memory Patterns
|
||||
schema: schema/research-project
|
||||
---
|
||||
```
|
||||
|
||||
Use when the note's `type` differs from the schema it should validate against, or when multiple schema variants exist.
|
||||
|
||||
**3. Implicit by type** — no `schema` field, resolved by matching `type`:
|
||||
|
||||
```yaml
|
||||
---
|
||||
title: Paul Graham
|
||||
type: Person
|
||||
---
|
||||
```
|
||||
|
||||
The system looks up a schema note where `entity: Person`. If found, it applies. If not, no validation occurs.
|
||||
|
||||
**4. No schema** — perfectly fine. Most notes don't need one.
|
||||
|
||||
### Schema Notes
|
||||
|
||||
A schema is itself a Basic Memory note with `type: schema`. It lives anywhere (though `schema/` is the conventional directory).
|
||||
|
||||
```yaml
|
||||
# schema/Person.md
|
||||
---
|
||||
title: Person
|
||||
type: schema
|
||||
entity: Person
|
||||
version: 1
|
||||
schema:
|
||||
name: string, full name
|
||||
role?: string, job title or position
|
||||
works_at?: Organization, employer
|
||||
expertise?(array): string, areas of knowledge
|
||||
email?: string, contact email
|
||||
settings:
|
||||
validation: warn
|
||||
---
|
||||
|
||||
# Person
|
||||
|
||||
A human individual in the knowledge graph.
|
||||
```
|
||||
|
||||
| Field | Required | Description |
|
||||
|-------|----------|-------------|
|
||||
| `type` | Yes | Must be `schema` |
|
||||
| `entity` | Yes | The entity type this schema describes (e.g., `Person`) |
|
||||
| `version` | No | Schema version number (default: `1`) |
|
||||
| `schema` | Yes | Picoschema dict defining the fields |
|
||||
| `settings.validation` | No | Validation mode (default: `warn`) |
|
||||
|
||||
Schema notes are regular notes — they show up in search, can have observations and relations, and participate in the knowledge graph.
|
||||
|
||||
### Validation Modes
|
||||
|
||||
| Mode | Behavior |
|
||||
|------|----------|
|
||||
| `warn` | Warnings in output, doesn't block (default) |
|
||||
| `strict` | Errors that block sync, for CI/CD enforcement |
|
||||
| `off` | No validation |
|
||||
|
||||
### Validation Output
|
||||
|
||||
```
|
||||
$ bm schema validate people/ada-lovelace.md
|
||||
|
||||
⚠ Person schema validation:
|
||||
- Missing required field: name (expected [name] observation)
|
||||
- Missing optional field: role
|
||||
- Missing optional field: works_at (no relation found)
|
||||
|
||||
ℹ Unmatched observations: [fact] ×2, [born] ×1
|
||||
ℹ Unmatched relations: collaborated_with
|
||||
```
|
||||
|
||||
"Unmatched" items are informational — observations and relations the schema doesn't cover.
|
||||
|
||||
### Schema Inference
|
||||
|
||||
Generate schemas from existing notes by analyzing observation and relation frequency:
|
||||
|
||||
```
|
||||
$ bm schema infer Person
|
||||
|
||||
Analyzing 30 notes with type: Person...
|
||||
|
||||
Observations found:
|
||||
[name] 30/30 100% → name: string
|
||||
[role] 27/30 90% → role?: string
|
||||
[expertise] 18/30 60% → expertise?(array): string
|
||||
[email] 8/30 27% → email?: string
|
||||
|
||||
Relations found:
|
||||
works_at 22/30 73% → works_at?: Organization
|
||||
|
||||
Suggested schema:
|
||||
name: string, full name
|
||||
role?: string, job title
|
||||
expertise?(array): string, areas of knowledge
|
||||
email?: string, contact email
|
||||
works_at?: Organization, employer
|
||||
|
||||
Save to schema/Person.md? [y/n]
|
||||
```
|
||||
|
||||
Frequency thresholds:
|
||||
- **100% present** → required field
|
||||
- **25%+ present** → optional field
|
||||
- **Below 25%** → excluded from suggestion
|
||||
|
||||
### Schema Drift Detection
|
||||
|
||||
Track how usage patterns shift over time:
|
||||
|
||||
```
|
||||
$ bm schema diff Person
|
||||
|
||||
Schema drift detected:
|
||||
|
||||
+ expertise: now in 81% of notes (was 12%)
|
||||
- department: dropped to 3% of notes
|
||||
~ works_at: cardinality changed (one → many)
|
||||
|
||||
Update schema? [y/n/review]
|
||||
```
|
||||
|
||||
## Complete Examples
|
||||
|
||||
### Simple Note (No Schema)
|
||||
|
||||
```markdown
|
||||
---
|
||||
title: Project Ideas
|
||||
type: note
|
||||
tags: [ideas, brainstorm]
|
||||
---
|
||||
|
||||
# Project Ideas
|
||||
|
||||
## Observations
|
||||
- [idea] Build a CLI tool for markdown linting #tooling
|
||||
- [idea] Create a recipe knowledge base #cooking
|
||||
- [priority] Focus on developer tools first (Q1 goal)
|
||||
|
||||
## Relations
|
||||
- inspired_by [[Developer Workflow Research]]
|
||||
- part_of [[Q1 Planning]]
|
||||
```
|
||||
|
||||
### Schema-Validated Note
|
||||
|
||||
Schema at `schema/Person.md`:
|
||||
|
||||
```yaml
|
||||
---
|
||||
title: Person
|
||||
type: schema
|
||||
entity: Person
|
||||
version: 1
|
||||
schema:
|
||||
name: string, full name
|
||||
role?: string, job title or position
|
||||
works_at?: Organization, employer
|
||||
expertise?(array): string, areas of knowledge
|
||||
email?: string, contact email
|
||||
settings:
|
||||
validation: warn
|
||||
---
|
||||
|
||||
# Person
|
||||
|
||||
A human individual in the knowledge graph.
|
||||
```
|
||||
|
||||
Note at `people/paul-graham.md`:
|
||||
|
||||
```markdown
|
||||
---
|
||||
title: Paul Graham
|
||||
type: Person
|
||||
tags: [startups, essays, lisp]
|
||||
---
|
||||
|
||||
# Paul Graham
|
||||
|
||||
## Observations
|
||||
- [name] Paul Graham
|
||||
- [role] Essayist and investor
|
||||
- [expertise] Startups
|
||||
- [expertise] Lisp
|
||||
- [expertise] Essay writing
|
||||
- [fact] Created Viaweb, the first web app
|
||||
|
||||
## Relations
|
||||
- works_at [[Y Combinator]]
|
||||
- authored [[Hackers and Painters]]
|
||||
```
|
||||
|
||||
The `[fact]` observation and `authored` relation are not in the schema — they're valid, just unmatched. The schema only checks that `[name]` exists (required) and looks for optional fields like `[role]`, `[expertise]`, and `works_at`.
|
||||
|
||||
### Inline Schema Note
|
||||
|
||||
```markdown
|
||||
---
|
||||
title: Team Standup 2024-01-15
|
||||
type: meeting
|
||||
schema:
|
||||
attendees(array): string, who was there
|
||||
decisions(array): string, what was decided
|
||||
action_items(array): string, follow-ups
|
||||
blockers?(array): string, anything stuck
|
||||
---
|
||||
|
||||
# Team Standup 2024-01-15
|
||||
|
||||
## Observations
|
||||
- [attendees] Paul
|
||||
- [attendees] Sarah
|
||||
- [decisions] Ship v2 by Friday
|
||||
- [action_items] Paul to review PR #42
|
||||
- [blockers] Waiting on API credentials
|
||||
```
|
||||
@@ -1,147 +0,0 @@
|
||||
# Simplified Local/Cloud Routing
|
||||
|
||||
## Context
|
||||
|
||||
Basic Memory now uses explicit, project-aware routing without a global cloud-mode toggle.
|
||||
Routing is determined by command-level flags and project mode, not by a global `cloud_mode` state.
|
||||
|
||||
This document is the canonical contract for local/cloud routing behavior in CLI, MCP, and API-adjacent clients.
|
||||
|
||||
## Goals
|
||||
|
||||
1. Remove global `cloud_mode` from runtime/routing semantics.
|
||||
2. Keep MCP HTTP/SSE local-only; let stdio honor per-project routing.
|
||||
3. Make CLI routing explicit and easy to reason about.
|
||||
4. Support projects that exist in both local and cloud without ambiguity.
|
||||
|
||||
## Routing Contract
|
||||
|
||||
Routing is resolved in this order:
|
||||
|
||||
1. Injected client factory (for composition/integration contexts)
|
||||
2. Explicit routing override (`--local` / `--cloud` or env vars below)
|
||||
3. Project-scoped routing (`project.mode`) when a project is known
|
||||
4. Default local routing
|
||||
|
||||
### Routing Environment Variables
|
||||
|
||||
- `BASIC_MEMORY_FORCE_LOCAL=true`: force local transport
|
||||
- `BASIC_MEMORY_FORCE_CLOUD=true`: force cloud proxy transport
|
||||
- `BASIC_MEMORY_EXPLICIT_ROUTING=true`: marks routing as explicitly chosen for this command
|
||||
|
||||
When explicit routing is active, project mode does not override the selected route.
|
||||
|
||||
## Config Semantics
|
||||
|
||||
- `project.mode` is the only config-based routing signal for project-scoped operations.
|
||||
- Legacy `cloud_mode` values may be encountered during migration/loading but are not used for routing behavior.
|
||||
- Normalization saves remove stale `cloud_mode` from `~/.basic-memory/config.json`.
|
||||
|
||||
### Example Config
|
||||
|
||||
```json
|
||||
{
|
||||
"projects": {
|
||||
"main": {
|
||||
"path": "/Users/me/basic-memory",
|
||||
"mode": "local",
|
||||
"local_sync_path": null,
|
||||
"bisync_initialized": false,
|
||||
"last_sync": null
|
||||
},
|
||||
"specs": {
|
||||
"path": "specs",
|
||||
"mode": "cloud",
|
||||
"local_sync_path": "/Users/me/dev/specs",
|
||||
"bisync_initialized": true,
|
||||
"last_sync": "2026-02-06T17:36:38.544153"
|
||||
}
|
||||
},
|
||||
"default_project": "main",
|
||||
"cloud_api_key": "bmc_abc123...",
|
||||
"cloud_host": "https://cloud.basicmemory.com"
|
||||
}
|
||||
```
|
||||
|
||||
## Cloud Commands Are Auth-Only
|
||||
|
||||
`bm cloud login`, `bm cloud logout`, and `bm cloud status` manage authentication state.
|
||||
|
||||
- `bm cloud login`
|
||||
- performs OAuth device flow
|
||||
- stores/refreshes token material
|
||||
- may verify cloud health/subscription
|
||||
- does not change routing defaults
|
||||
- `bm cloud logout`
|
||||
- removes stored OAuth session tokens
|
||||
- does not change routing defaults
|
||||
- `bm cloud status`
|
||||
- reports auth state (API key, OAuth token validity)
|
||||
- runs health checks only when credentials are available
|
||||
|
||||
## MCP Transport Routing
|
||||
|
||||
### Stdio (default)
|
||||
|
||||
`bm mcp --transport stdio` uses natural per-project routing.
|
||||
|
||||
- Local-mode projects route through the in-process ASGI transport.
|
||||
- Cloud-mode projects route to the cloud proxy with Bearer auth (API key).
|
||||
- No explicit routing env vars are injected by the CLI command.
|
||||
- Externally-set env vars are honored (e.g. `BASIC_MEMORY_FORCE_CLOUD=true` for cloud deployments).
|
||||
- Users who need all projects forced local can set `BASIC_MEMORY_FORCE_LOCAL=true` externally.
|
||||
|
||||
### HTTP and SSE Transports
|
||||
|
||||
`bm mcp --transport streamable-http` and `bm mcp --transport sse` always route locally.
|
||||
|
||||
These transports set explicit local routing (`BASIC_MEMORY_FORCE_LOCAL=true` and
|
||||
`BASIC_MEMORY_EXPLICIT_ROUTING=true`) before starting the server. This prevents cloud
|
||||
routing regardless of project mode, since HTTP/SSE serve as local API endpoints.
|
||||
|
||||
## Project List UX for Dual Presence
|
||||
|
||||
Projects may exist in both local and cloud. `bm project list` should display that clearly in one row per logical
|
||||
project identity, with explicit source/target signals.
|
||||
|
||||
Recommended display contract:
|
||||
|
||||
1. Keep one row per normalized project name/permalink.
|
||||
2. Show both local and cloud presence as separate columns/indicators.
|
||||
3. Show an explicit `MCP (stdio)` target column that always resolves to `local`.
|
||||
4. Keep CLI route semantics explicit:
|
||||
- no flags: default local for non-project commands
|
||||
- `--cloud`: force cloud
|
||||
- `--local`: force local
|
||||
|
||||
## Project LS Targeting
|
||||
|
||||
`bm project ls` should clearly identify which project instance is being listed.
|
||||
|
||||
Targeting rules:
|
||||
|
||||
1. No routing flags: list local project files.
|
||||
2. `--cloud`: list cloud project files.
|
||||
3. `--local`: list local project files (explicit override).
|
||||
4. Output should label the active target (`LOCAL` or `CLOUD`) in heading or status line.
|
||||
|
||||
## Runtime Mode
|
||||
|
||||
Runtime mode is no longer a cloud/local routing switch for local app flows.
|
||||
|
||||
- `resolve_runtime_mode(is_test_env)` resolves to:
|
||||
- `TEST` when running in test environment
|
||||
- `LOCAL` otherwise
|
||||
- `RuntimeMode.CLOUD` may remain for compatibility with existing tests/call sites but is not selected by normal local
|
||||
runtime resolution.
|
||||
|
||||
## Verification Checklist
|
||||
|
||||
1. Loading config with legacy `cloud_mode` succeeds.
|
||||
2. Saving config strips legacy `cloud_mode`.
|
||||
3. `--local/--cloud` always override per-project mode for that command.
|
||||
4. No-project + no-flags commands route local by default.
|
||||
5. `bm cloud login/logout` do not toggle routing behavior.
|
||||
6. `bm mcp` stdio routes per-project mode; HTTP/SSE remain local-forced.
|
||||
7. `bm project list` communicates dual local/cloud presence without ambiguity.
|
||||
8. `bm project ls` output identifies route target explicitly.
|
||||
@@ -91,11 +91,10 @@ SQLite Database (Index)
|
||||
# List all projects
|
||||
projects = await list_memory_projects()
|
||||
|
||||
# Response structure (each entry includes external_id you can pass as project_id):
|
||||
# Response structure:
|
||||
# [
|
||||
# {
|
||||
# "name": "main",
|
||||
# "external_id": "550e8400-e29b-41d4-a716-446655440000",
|
||||
# "path": "/Users/name/notes",
|
||||
# "is_default": True,
|
||||
# "note_count": 156,
|
||||
@@ -103,7 +102,6 @@ projects = await list_memory_projects()
|
||||
# },
|
||||
# {
|
||||
# "name": "work",
|
||||
# "external_id": "9f86d081-884c-42a3-b5e3-1c0c5b4c8e52",
|
||||
# "path": "/Users/name/work-notes",
|
||||
# "is_default": False,
|
||||
# "note_count": 89,
|
||||
@@ -166,44 +164,6 @@ active_project = "main"
|
||||
results = await search_notes(query="topic", project=active_project)
|
||||
```
|
||||
|
||||
### `project` vs `project_id`
|
||||
|
||||
Every project has two identifiers:
|
||||
|
||||
- **`project`** — human-readable name (e.g., `"main"`). Easy to use, but can collide across cloud workspaces.
|
||||
- **`project_id`** — stable `external_id` UUID. Always unambiguous; takes precedence over `project` when both are passed.
|
||||
|
||||
**When to prefer `project_id`:**
|
||||
|
||||
1. **Cloud multi-workspace setups.** If the user belongs to more than one workspace (personal + organization, or several organizations) and the same project name might exist in more than one of them, pass `project_id` to route to the exact project. Without it, name resolution falls back to the default workspace, which may not be the one the user means.
|
||||
2. **After `list_memory_projects()`.** Once you have the `external_id`, prefer using it — it's the same number of characters in JSON and saves a name-resolution round-trip.
|
||||
3. **When persisting a project choice across a long session.** UUIDs are stable; names can be renamed.
|
||||
|
||||
**When `project` (name) is fine:**
|
||||
|
||||
- Local single-workspace setups (no collision risk).
|
||||
- One-off operations where the name is clearly visible to the user (e.g., quick `search_notes(project="main", ...)`).
|
||||
- The user explicitly references a project by name in their message.
|
||||
|
||||
**Example — cloud multi-workspace pattern:**
|
||||
|
||||
```python
|
||||
# Discover and pick the right project for this user
|
||||
projects = await list_memory_projects()
|
||||
target = next(p for p in projects if p["name"] == "research" and p["workspace"]["slug"] == "acme")
|
||||
|
||||
# Use the UUID for all subsequent operations — no ambiguity
|
||||
await write_note(
|
||||
title="Meeting Notes",
|
||||
content="...",
|
||||
folder="meetings",
|
||||
project_id=target["external_id"],
|
||||
)
|
||||
results = await search_notes(query="kickoff", project_id=target["external_id"])
|
||||
```
|
||||
|
||||
**Precedence rule:** When both are passed, `project_id` wins. This lets you safely supply `project="main"` for backward compatibility while still routing precisely with `project_id`.
|
||||
|
||||
### Cross-Project Operations
|
||||
|
||||
**Some tools work across all projects when project parameter omitted:**
|
||||
@@ -467,8 +427,6 @@ await write_note(
|
||||
)
|
||||
```
|
||||
|
||||
> **Important**: `write_note` errors if the note already exists. Use `edit_note` for incremental changes, or pass `overwrite=True` to replace.
|
||||
|
||||
**Well-structured note**:
|
||||
|
||||
```python
|
||||
@@ -802,9 +760,6 @@ notes = await read_note(
|
||||
identifier="memory://specs/*",
|
||||
project="main"
|
||||
)
|
||||
|
||||
# Cross-project URL (auto-routes to the correct project)
|
||||
note = await read_note(identifier="memory://research/specs/api-design")
|
||||
```
|
||||
|
||||
```python
|
||||
@@ -1083,41 +1038,9 @@ recent_decisions = await search_notes(
|
||||
)
|
||||
```
|
||||
|
||||
**Structured frontmatter filters**:
|
||||
|
||||
```python
|
||||
# Filter by tags and status
|
||||
results = await search_notes(
|
||||
query="authentication",
|
||||
tags=["security"],
|
||||
status="in-progress",
|
||||
project="main"
|
||||
)
|
||||
|
||||
# Complex metadata filters (supports $in, $gt, $gte, $lt, $lte, $between)
|
||||
results = await search_notes(
|
||||
query="api design",
|
||||
metadata_filters={
|
||||
"type": "spec",
|
||||
"priority": {"$in": ["high", "critical"]},
|
||||
"tags": ["architecture"]
|
||||
},
|
||||
project="main"
|
||||
)
|
||||
|
||||
# Metadata-only search (no query needed)
|
||||
results = await search_notes(
|
||||
metadata_filters={"type": "spec", "status": "in-progress"},
|
||||
project="main"
|
||||
)
|
||||
```
|
||||
|
||||
### Search Types
|
||||
|
||||
Available types: `"text"`, `"title"`, `"permalink"`, `"vector"`/`"semantic"`, `"hybrid"`.
|
||||
Default is `"hybrid"` when semantic search is enabled, `"text"` otherwise.
|
||||
|
||||
**Text search**:
|
||||
**Text search (default)**:
|
||||
|
||||
```python
|
||||
# Full-text search across all content
|
||||
@@ -1128,52 +1051,17 @@ results = await search_notes(
|
||||
)
|
||||
```
|
||||
|
||||
**Title and permalink search**:
|
||||
|
||||
```python
|
||||
# Search by title only
|
||||
results = await search_notes(query="API Design", search_type="title", project="main")
|
||||
|
||||
# Search by permalink
|
||||
results = await search_notes(query="specs/api-design", search_type="permalink", project="main")
|
||||
```
|
||||
|
||||
**Semantic/vector search**:
|
||||
**Semantic search**:
|
||||
|
||||
```python
|
||||
# Semantic/vector search (if enabled)
|
||||
results = await search_notes(
|
||||
query="user login security",
|
||||
search_type="semantic", # or "vector"
|
||||
project="main"
|
||||
)
|
||||
|
||||
# Override similarity threshold
|
||||
results = await search_notes(
|
||||
query="user login security",
|
||||
search_type="semantic",
|
||||
min_similarity=0.5,
|
||||
project="main"
|
||||
)
|
||||
```
|
||||
|
||||
**Hybrid search** (combines text + semantic):
|
||||
|
||||
```python
|
||||
results = await search_notes(
|
||||
query="authentication best practices",
|
||||
search_type="hybrid",
|
||||
project="main"
|
||||
)
|
||||
```
|
||||
|
||||
**Tag shorthand in query**:
|
||||
|
||||
```python
|
||||
# Use tag: prefix as shorthand
|
||||
results = await search_notes(query="tag:security", project="main")
|
||||
```
|
||||
|
||||
### Search Response
|
||||
|
||||
**Result structure**:
|
||||
@@ -2244,31 +2132,6 @@ active_project = projects[0]["name"]
|
||||
results = await search_notes(query="test", project=active_project)
|
||||
```
|
||||
|
||||
### Note Already Exists
|
||||
|
||||
**Error**: `write_note` called for a note that already exists
|
||||
|
||||
**Solution**:
|
||||
|
||||
```python
|
||||
# Preferred: use edit_note for incremental updates
|
||||
await edit_note(
|
||||
identifier="Existing Topic",
|
||||
operation="append",
|
||||
content="\n- [update] new information",
|
||||
project="main"
|
||||
)
|
||||
|
||||
# Alternative: replace the entire note
|
||||
await write_note(
|
||||
title="Existing Topic",
|
||||
content="# Existing Topic\n...",
|
||||
folder="notes",
|
||||
overwrite=True,
|
||||
project="main"
|
||||
)
|
||||
```
|
||||
|
||||
### Entity Not Found
|
||||
|
||||
**Error**: Note doesn't exist
|
||||
@@ -2824,15 +2687,14 @@ await write_note(
|
||||
|
||||
### Content Management
|
||||
|
||||
**write_note(title, content, folder, tags, note_type, overwrite, project)**
|
||||
- Create new markdown notes (errors if note already exists unless overwrite=True)
|
||||
**write_note(title, content, folder, tags, note_type, project)**
|
||||
- Create or update markdown notes
|
||||
- Parameters:
|
||||
- `title` (required): Note title
|
||||
- `content` (required): Markdown content
|
||||
- `folder` (required): Destination folder
|
||||
- `tags` (optional): List of tags
|
||||
- `note_type` (optional): Type of note (stored in frontmatter). Can be "note", "person", "meeting", "guide", etc.
|
||||
- `overwrite` (optional): Set to True to replace an existing note (default: error if exists)
|
||||
- `project` (required unless default_project_mode): Target project
|
||||
- Returns: Created/updated entity with permalink
|
||||
- Example:
|
||||
@@ -2999,20 +2861,16 @@ contents = await list_directory(
|
||||
|
||||
### Search & Discovery
|
||||
|
||||
**search_notes(query, page, page_size, search_type, types, entity_types, after_date, metadata_filters, tags, status, min_similarity, project)**
|
||||
**search_notes(query, page, page_size, search_type, types, entity_types, after_date, project)**
|
||||
- Search across knowledge base
|
||||
- Parameters:
|
||||
- `query` (optional): Search query (not required for filter-only searches)
|
||||
- `query` (required): Search query
|
||||
- `page` (optional): Page number (default: 1)
|
||||
- `page_size` (optional): Results per page (default: 10)
|
||||
- `search_type` (optional): "text", "title", "permalink", "vector"/"semantic", "hybrid" (default: "hybrid" when semantic enabled, "text" otherwise)
|
||||
- `search_type` (optional): "text" or "semantic"
|
||||
- `types` (optional): Entity type filter
|
||||
- `entity_types` (optional): Observation category filter
|
||||
- `after_date` (optional): Date filter (ISO format)
|
||||
- `metadata_filters` (optional): Structured frontmatter filters (dict, supports `$in`, `$gt`, `$gte`, `$lt`, `$lte`, `$between` operators)
|
||||
- `tags` (optional): Frontmatter tags filter (list); also available via `tag:` query shorthand
|
||||
- `status` (optional): Frontmatter status filter (string)
|
||||
- `min_similarity` (optional): Override similarity threshold for vector/hybrid search
|
||||
- `project` (required unless default_project_mode): Target project
|
||||
- Returns: Matching entities with scores
|
||||
- Example:
|
||||
@@ -3025,15 +2883,6 @@ results = await search_notes(
|
||||
)
|
||||
```
|
||||
|
||||
**Metadata-only search (via search_notes)**
|
||||
- Use `search_notes` with `metadata_filters` and no `query` for metadata-only searches:
|
||||
```python
|
||||
results = await search_notes(
|
||||
metadata_filters={"type": "spec", "status": "in-progress"},
|
||||
project="main"
|
||||
)
|
||||
```
|
||||
|
||||
### Project Management
|
||||
|
||||
**list_memory_projects()**
|
||||
@@ -3081,15 +2930,6 @@ await delete_project(project_name="old-project")
|
||||
status = await sync_status(project="main")
|
||||
```
|
||||
|
||||
**list_workspaces()**
|
||||
- List available workspaces (cloud)
|
||||
- Parameters: None
|
||||
- Returns: List of workspaces with metadata
|
||||
- Example:
|
||||
```python
|
||||
workspaces = await list_workspaces()
|
||||
```
|
||||
|
||||
### Visualization
|
||||
|
||||
**canvas(nodes, edges, title, folder, project)**
|
||||
@@ -3358,8 +3198,8 @@ await edit_note(
|
||||
project="main"
|
||||
)
|
||||
|
||||
# When full rewrite is needed, use overwrite=True
|
||||
await write_note(title="Note", content="...", folder="notes", overwrite=True)
|
||||
# Avoid: Complete rewrite
|
||||
# (unless necessary for major restructuring)
|
||||
```
|
||||
|
||||
### 14. Tagging Strategy
|
||||
|
||||
@@ -184,8 +184,8 @@ finance/ (lowercase f)
|
||||
Use Basic Memory's built-in conflict detection:
|
||||
|
||||
```bash
|
||||
# Index local file changes (conflicts are handled during the scan)
|
||||
basic-memory reindex
|
||||
# Sync will report conflicts
|
||||
basic-memory sync
|
||||
|
||||
# Check sync status for warnings
|
||||
basic-memory status
|
||||
|
||||
+140
-460
@@ -5,36 +5,18 @@ The Basic Memory Cloud CLI provides seamless integration between local and cloud
|
||||
## Overview
|
||||
|
||||
The cloud CLI enables you to:
|
||||
- **Authenticate cloud access** - OAuth/API key credentials are stored locally for cloud operations
|
||||
- **Toggle cloud mode** - All regular `bm` commands work with cloud when enabled
|
||||
- **Project-scoped sync** - Each project independently manages its sync configuration
|
||||
- **Explicit operations** - Sync only what you want, when you want
|
||||
- **Team-safe push/pull** - Additive, git-style transfers that work on shared Team workspaces
|
||||
- **Bidirectional sync** - Keep local and cloud in sync with rclone bisync (Personal workspaces)
|
||||
- **Bidirectional sync** - Keep local and cloud in sync with rclone bisync
|
||||
- **Offline access** - Work locally, sync when ready
|
||||
|
||||
### Personal vs Team workspaces
|
||||
|
||||
The transfer commands fall into two groups:
|
||||
|
||||
| Command | Direction | Behavior | Personal | Team |
|
||||
|---|---|---|---|---|
|
||||
| `bm cloud pull` | cloud → local | **additive** — never deletes local | ✅ | ✅ |
|
||||
| `bm cloud push` | local → cloud | **additive** — never deletes cloud | ✅ | ✅ |
|
||||
| `bm cloud sync` | local → cloud | **mirror** — deletes cloud files missing locally | ✅ | ❌ |
|
||||
| `bm cloud bisync` | local ↔ cloud | **mirror** — two-way, deletes on both sides | ✅ | ❌ |
|
||||
|
||||
`sync` and `bisync` are mirror operations: one local tree becomes authoritative and files missing on the other side get deleted. That is correct for a Personal workspace (one user, one source of truth) but unsafe on a shared Team bucket, where it could delete a teammate's files. On Team workspaces these commands exit early with a clear error and point you at `push`/`pull`.
|
||||
|
||||
`push` and `pull` are additive (they use `rclone copy`, which never deletes on the destination), so they are safe on both Personal and Team workspaces.
|
||||
|
||||
## Prerequisites
|
||||
|
||||
Before using Basic Memory Cloud, you need:
|
||||
|
||||
- **Active Subscription**: An active Basic Memory Cloud subscription is required to access cloud features
|
||||
- **Subscribe**: Visit [https://basicmemory.com/subscribe](https://basicmemory.com/subscribe) to sign up
|
||||
- **Optional**: Cloud is optional. Local-first open-source usage continues without cloud.
|
||||
- **OSS Discount**: Use code `{{OSS_DISCOUNT_CODE}}` for 20% off for 3 months.
|
||||
|
||||
If you attempt to log in without an active subscription, you'll receive a "Subscription Required" error with a link to subscribe.
|
||||
|
||||
@@ -56,7 +38,7 @@ If you attempt to log in without an active subscription, you'll receive a "Subsc
|
||||
|
||||
1. **Cloud-only** - Project exists on cloud, no local copy
|
||||
2. **Cloud + Local (synced)** - Project has a local working directory that syncs
|
||||
3. **Local-only** - Project exists locally and is not routed to cloud
|
||||
3. **Local-only** - Project exists locally (when cloud mode is disabled)
|
||||
|
||||
**Example:**
|
||||
|
||||
@@ -66,13 +48,13 @@ If you attempt to log in without an active subscription, you'll receive a "Subsc
|
||||
# - work: wants local sync at ~/work-notes
|
||||
# - temp: cloud-only, no local sync needed
|
||||
|
||||
bm project add research --cloud --local-path ~/Documents/research
|
||||
bm project add work --cloud --local-path ~/work-notes
|
||||
bm project add temp --cloud # No local sync
|
||||
bm project add research --local-path ~/Documents/research
|
||||
bm project add work --local-path ~/work-notes
|
||||
bm project add temp # No local sync
|
||||
|
||||
# Now you can sync individually (after initial --resync):
|
||||
bm cloud bisync --name research
|
||||
bm cloud bisync --name work
|
||||
bm project bisync --name research
|
||||
bm project bisync --name work
|
||||
# temp stays cloud-only
|
||||
```
|
||||
|
||||
@@ -84,9 +66,9 @@ bm cloud bisync --name work
|
||||
|
||||
## Quick Start
|
||||
|
||||
### 1. Authenticate Cloud Access
|
||||
### 1. Enable Cloud Mode
|
||||
|
||||
Authenticate with cloud:
|
||||
Authenticate and enable cloud mode:
|
||||
|
||||
```bash
|
||||
bm cloud login
|
||||
@@ -94,12 +76,11 @@ bm cloud login
|
||||
|
||||
**What this does:**
|
||||
1. Opens browser to Basic Memory Cloud authentication page
|
||||
2. Stores authentication tokens in `~/.basic-memory/basic-memory-cloud.json`
|
||||
3. Validates your subscription status
|
||||
4. Leaves routing behavior unchanged (auth only)
|
||||
2. Stores authentication token in `~/.basic-memory/auth/token`
|
||||
3. **Enables cloud mode** - all CLI commands now work against cloud
|
||||
4. Validates your subscription status
|
||||
|
||||
**Result:** Cloud credentials are available for cloud-routed commands.
|
||||
Apply OSS discount code `{{OSS_DISCOUNT_CODE}}` during checkout to receive 20% off for 3 months.
|
||||
**Result:** All `bm project`, `bm tools` commands now work with cloud.
|
||||
|
||||
### 2. Set Up Sync
|
||||
|
||||
@@ -110,38 +91,33 @@ bm cloud setup
|
||||
```
|
||||
|
||||
**What this does:**
|
||||
1. Installs rclone with a supported package manager (if needed)
|
||||
1. Installs rclone automatically (if needed)
|
||||
2. Fetches your tenant information from cloud
|
||||
3. Generates scoped S3 credentials for sync
|
||||
4. Configures single rclone remote: `basic-memory-cloud`
|
||||
|
||||
**Result:** You're ready to sync projects. No sync directories created yet - those come with project setup.
|
||||
|
||||
Rclone setup uses package managers such as Homebrew, MacPorts, apt, dnf, yum, pacman,
|
||||
zypper, snap, winget, Chocolatey, or Scoop when available. It does not run remote
|
||||
install scripts with `sudo`; if no supported package manager is found, the CLI prints
|
||||
manual install instructions.
|
||||
|
||||
### 3. Add Projects with Sync
|
||||
|
||||
Create projects with optional local sync paths:
|
||||
|
||||
```bash
|
||||
# Create cloud project without local sync
|
||||
bm project add research --cloud
|
||||
bm project add research
|
||||
|
||||
# Create cloud project WITH local sync
|
||||
bm project add research --cloud --local-path ~/Documents/research
|
||||
bm project add research --local-path ~/Documents/research
|
||||
|
||||
# Or configure sync for existing project
|
||||
bm cloud sync-setup research ~/Documents/research
|
||||
bm project sync-setup research ~/Documents/research
|
||||
```
|
||||
|
||||
**What happens under the covers:**
|
||||
|
||||
When you add a project with `--local-path`:
|
||||
1. Project created on cloud at `/app/data/research`
|
||||
2. Local path stored in config for that project (`local_sync_path`)
|
||||
2. Local path stored in config: `cloud_projects.research.local_path = "~/Documents/research"`
|
||||
3. Local directory created if it doesn't exist
|
||||
4. Bisync state directory created at `~/.basic-memory/bisync-state/research/`
|
||||
|
||||
@@ -153,10 +129,10 @@ Establish the initial sync baseline. **Best practice:** Always preview with `--d
|
||||
|
||||
```bash
|
||||
# Step 1: Preview the initial sync (recommended)
|
||||
bm cloud bisync --name research --resync --dry-run
|
||||
bm project bisync --name research --resync --dry-run
|
||||
|
||||
# Step 2: If all looks good, run the actual sync
|
||||
bm cloud bisync --name research --resync
|
||||
bm project bisync --name research --resync
|
||||
```
|
||||
|
||||
**What happens under the covers:**
|
||||
@@ -183,7 +159,7 @@ This will effectively make both Path1 and Path2 filesystems contain a matching s
|
||||
After the first sync, just run bisync without `--resync`:
|
||||
|
||||
```bash
|
||||
bm cloud bisync --name research
|
||||
bm project bisync --name research
|
||||
```
|
||||
|
||||
**What happens:**
|
||||
@@ -203,8 +179,7 @@ bm cloud status
|
||||
```
|
||||
|
||||
You should see:
|
||||
- `OAuth: token valid` (or missing/expired)
|
||||
- `API Key: configured` (or not set)
|
||||
- `Mode: Cloud (enabled)`
|
||||
- `Cloud instance is healthy`
|
||||
- Instructions for project sync commands
|
||||
|
||||
@@ -212,16 +187,16 @@ You should see:
|
||||
|
||||
### Understanding Project Commands
|
||||
|
||||
**Key concept:** Use regular `bm project` commands (not `bm cloud project`).
|
||||
**Key concept:** When cloud mode is enabled, use regular `bm project` commands (not `bm cloud project`).
|
||||
|
||||
```bash
|
||||
# Local route
|
||||
bm project list --local
|
||||
bm project add research ~/Documents/research
|
||||
# In cloud mode:
|
||||
bm project list # Lists cloud projects
|
||||
bm project add research # Creates cloud project
|
||||
|
||||
# Cloud route
|
||||
bm project list --cloud
|
||||
bm project add research --cloud
|
||||
# In local mode:
|
||||
bm project list # Lists local projects
|
||||
bm project add research ~/Documents/research # Creates local project
|
||||
```
|
||||
|
||||
### Creating Projects
|
||||
@@ -229,7 +204,7 @@ bm project add research --cloud
|
||||
**Use case 1: Cloud-only project (no local sync)**
|
||||
|
||||
```bash
|
||||
bm project add temp-notes --cloud
|
||||
bm project add temp-notes
|
||||
```
|
||||
|
||||
**What this does:**
|
||||
@@ -242,7 +217,7 @@ bm project add temp-notes --cloud
|
||||
**Use case 2: Cloud project with local sync**
|
||||
|
||||
```bash
|
||||
bm project add research --cloud --local-path ~/Documents/research
|
||||
bm project add research --local-path ~/Documents/research
|
||||
```
|
||||
|
||||
**What this does:**
|
||||
@@ -251,13 +226,13 @@ bm project add research --cloud --local-path ~/Documents/research
|
||||
- Stores sync config in `~/.basic-memory/config.json`
|
||||
- Prepares for bisync (but doesn't sync yet)
|
||||
|
||||
**Result:** Project ready to sync. Run `bm cloud bisync --name research --resync` to establish baseline.
|
||||
**Result:** Project ready to sync. Run `bm project bisync --name research --resync` to establish baseline.
|
||||
|
||||
**Use case 3: Add sync to existing cloud project**
|
||||
|
||||
```bash
|
||||
# Project already exists on cloud
|
||||
bm cloud sync-setup research ~/Documents/research
|
||||
bm project sync-setup research ~/Documents/research
|
||||
```
|
||||
|
||||
**What this does:**
|
||||
@@ -276,132 +251,28 @@ bm project list
|
||||
```
|
||||
|
||||
**What you see:**
|
||||
- Local projects always
|
||||
- Cloud projects when credentials are available
|
||||
- All projects in cloud (when cloud mode enabled)
|
||||
- Default project marked
|
||||
- Route-related metadata (for example, local/cloud presence and sync info)
|
||||
- Project paths shown
|
||||
|
||||
Example shape (single row for dual-presence projects):
|
||||
|
||||
```text
|
||||
Name Path Local Path Cloud Path CLI Default MCP (stdio)
|
||||
main /basic-memory ~/basic-memory /basic-memory local local
|
||||
specs /specs ~/dev/specs /specs cloud local
|
||||
```
|
||||
|
||||
### When a Project Exists in Both Local and Cloud
|
||||
|
||||
Use routing flags to disambiguate command targets:
|
||||
|
||||
```bash
|
||||
# Force local target for this command
|
||||
bm project info main --local
|
||||
bm project ls --name main --local
|
||||
|
||||
# Force cloud target for this command
|
||||
bm project info main --cloud
|
||||
bm project ls --name main --cloud
|
||||
```
|
||||
|
||||
Default behavior for no-project, no-flag commands is local.
|
||||
For MCP stdio, routing is always local.
|
||||
**Future:** Will show sync status (synced/not synced, last sync time).
|
||||
|
||||
## File Synchronization
|
||||
|
||||
### Understanding the Sync Commands
|
||||
|
||||
**There are five sync-related commands:**
|
||||
**There are three sync-related commands:**
|
||||
|
||||
| Command | Direction | Workspace | Summary |
|
||||
|---|---|---|---|
|
||||
| `bm cloud pull` | cloud → local | Personal + Team | Fetch cloud changes, additively (git-style) |
|
||||
| `bm cloud push` | local → cloud | Personal + Team | Upload local changes, additively (git-style) |
|
||||
| `bm cloud sync` | local → cloud | Personal only | One-way mirror (cloud becomes identical to local) |
|
||||
| `bm cloud bisync` | local ↔ cloud | Personal only | Two-way mirror (recommended for solo use) |
|
||||
| `bm cloud check` | — | Personal only | Verify mirror integrity (no changes) |
|
||||
1. `bm project sync` - One-way: local → cloud (make cloud match local)
|
||||
2. `bm project bisync` - Two-way: local ↔ cloud (recommended)
|
||||
3. `bm project check` - Verify files match (no changes)
|
||||
|
||||
If you collaborate on a shared Team workspace, use **`push`/`pull`** (see [Team Workspaces](#team-workspaces-push--pull-additive-git-style)). If you are the only writer (a Personal workspace), the mirror commands `sync`/`bisync` give you a single source of truth.
|
||||
|
||||
### Team Workspaces: push / pull (additive, git-style)
|
||||
|
||||
`push` and `pull` are the Team-safe transfer commands. They model `git push` / `git pull`:
|
||||
|
||||
- **`bm cloud pull`** fetches changes from the cloud into your local directory.
|
||||
- **`bm cloud push`** uploads your local changes to the cloud.
|
||||
|
||||
Both use `rclone copy`, so they are **additive — they never delete on the destination**. A conflict (a file that differs on both sides) is never resolved silently: by default the command aborts and lists the conflicting files, exactly like git refusing to clobber your changes.
|
||||
|
||||
#### Pull: fetch cloud changes
|
||||
|
||||
```bash
|
||||
# Preview first (recommended)
|
||||
bm cloud pull --name research --dry-run
|
||||
|
||||
# Fetch new/changed cloud files into local
|
||||
bm cloud pull --name research
|
||||
```
|
||||
|
||||
**What happens:**
|
||||
1. Compares cloud and local with `rclone check`
|
||||
2. Downloads files that are new or changed on the cloud
|
||||
3. Leaves your local-only files untouched (never deletes local)
|
||||
4. If any file differs on both sides, aborts and lists the conflicts (unless you pass `--on-conflict`)
|
||||
|
||||
#### Push: upload local changes
|
||||
|
||||
```bash
|
||||
bm cloud push --name research --dry-run
|
||||
bm cloud push --name research
|
||||
```
|
||||
|
||||
**What happens:**
|
||||
1. Compares local and cloud with `rclone check`
|
||||
2. Uploads files that are new or changed locally
|
||||
3. Leaves cloud-only files untouched (never deletes cloud)
|
||||
4. If any file differs on both sides, aborts and lists the conflicts — pull first, like a rejected `git push`
|
||||
|
||||
#### Resolving conflicts
|
||||
|
||||
When `push`/`pull` reports conflicts, re-run with `--on-conflict` to choose how differing files are handled. The value names exactly what survives, so it reads the same in both directions:
|
||||
|
||||
| `--on-conflict` | Behavior |
|
||||
|---|---|
|
||||
| `fail` *(default)* | List the conflicting files and exit without transferring anything |
|
||||
| `keep-cloud` | Take the cloud version (pull: overwrite local; push: skip those files) |
|
||||
| `keep-local` | Keep the local version (pull: skip those files; push: overwrite cloud) |
|
||||
| `keep-both` | Keep both — write the incoming version beside the existing one as `name.conflict-<date>.md` |
|
||||
|
||||
```bash
|
||||
# A teammate edited notes you also changed locally — pull reports a conflict:
|
||||
bm cloud pull --name research
|
||||
# pull aborted: 1 file(s) differ between local and cloud.
|
||||
# * notes/decisions.md
|
||||
# Re-run with one of: --on-conflict keep-cloud | keep-local | keep-both
|
||||
|
||||
# Take the cloud copy:
|
||||
bm cloud pull --name research --on-conflict keep-cloud
|
||||
|
||||
# Or keep both versions to merge by hand:
|
||||
bm cloud pull --name research --on-conflict keep-both
|
||||
```
|
||||
|
||||
#### Limitations
|
||||
|
||||
`push`/`pull` are deliberately simple, conflict-aware byte transfers — not a full reconciler. Without a sync baseline:
|
||||
|
||||
- **Deletions are not propagated.** A note deleted on one side is not removed from the other (we cannot tell an intentional delete from a file the other side never had). This is surfaced in the command output.
|
||||
- **Every divergence is treated as a conflict.** We cannot tell a teammate's edit from your stale copy, so any differing file prompts a decision rather than auto-resolving.
|
||||
|
||||
For conflict-aware *editing*, write through the MCP/API tools (which merge at the note level). A Team-safe bidirectional reconciler with a real baseline is tracked in [issue #862](https://github.com/basicmachines-co/basic-memory/issues/862).
|
||||
|
||||
### One-Way Sync: Local → Cloud (Personal only)
|
||||
### One-Way Sync: Local → Cloud
|
||||
|
||||
**Use case:** You made changes locally and want to push to cloud (overwrite cloud).
|
||||
|
||||
> **Personal workspaces only.** `sync` is a destructive mirror — it deletes cloud files that are not present locally. On a Team workspace it would delete a teammate's files, so it is blocked there. Use `bm cloud push` (additive) on Team workspaces.
|
||||
|
||||
```bash
|
||||
bm cloud sync --name research
|
||||
bm project sync --name research
|
||||
```
|
||||
|
||||
**What happens:**
|
||||
@@ -417,18 +288,16 @@ bm cloud sync --name research
|
||||
- You want to force cloud to match local
|
||||
- You don't care about cloud changes
|
||||
|
||||
### Two-Way Sync: Local ↔ Cloud (Personal only, recommended for solo use)
|
||||
### Two-Way Sync: Local ↔ Cloud (Recommended)
|
||||
|
||||
**Use case:** You edit files both locally and in cloud UI, want both to stay in sync.
|
||||
|
||||
> **Personal workspaces only.** `bisync` is a two-way mirror that can delete and overwrite on both sides. It is blocked on Team workspaces — use `bm cloud pull` then `bm cloud push` there. A Team-safe bidirectional reconciler is tracked separately ([issue #862](https://github.com/basicmachines-co/basic-memory/issues/862)).
|
||||
|
||||
```bash
|
||||
# First time - establish baseline
|
||||
bm cloud bisync --name research --resync
|
||||
bm project bisync --name research --resync
|
||||
|
||||
# Subsequent syncs
|
||||
bm cloud bisync --name research
|
||||
bm project bisync --name research
|
||||
```
|
||||
|
||||
**What happens:**
|
||||
@@ -447,7 +316,7 @@ echo "Local change" > ~/Documents/research/notes.md
|
||||
# Cloud now has: "Cloud change"
|
||||
|
||||
# Run bisync
|
||||
bm cloud bisync --name research
|
||||
bm project bisync --name research
|
||||
|
||||
# Result: Newer file wins (based on modification time)
|
||||
# If cloud was more recent, cloud version kept
|
||||
@@ -459,14 +328,12 @@ bm cloud bisync --name research
|
||||
- You edit in multiple places
|
||||
- You want automatic conflict resolution
|
||||
|
||||
### Verify Sync Integrity (Personal only)
|
||||
### Verify Sync Integrity
|
||||
|
||||
**Use case:** Check if local and cloud match without making changes.
|
||||
|
||||
> **Personal workspaces only.** `check` compares against the Personal workspace mirror remote, like `sync`/`bisync`. On Team workspaces use `bm cloud pull --dry-run` / `bm cloud push --dry-run` to preview differences instead.
|
||||
|
||||
```bash
|
||||
bm cloud check --name research
|
||||
bm project check --name research
|
||||
```
|
||||
|
||||
**What happens:**
|
||||
@@ -478,7 +345,7 @@ bm cloud check --name research
|
||||
|
||||
```bash
|
||||
# One-way check (faster)
|
||||
bm cloud check --name research --one-way
|
||||
bm project check --name research --one-way
|
||||
```
|
||||
|
||||
### Preview Changes (Dry Run)
|
||||
@@ -486,7 +353,7 @@ bm cloud check --name research --one-way
|
||||
**Use case:** See what would change without actually syncing.
|
||||
|
||||
```bash
|
||||
bm cloud bisync --name research --dry-run
|
||||
bm project bisync --name research --dry-run
|
||||
```
|
||||
|
||||
**What happens:**
|
||||
@@ -496,28 +363,24 @@ bm cloud bisync --name research --dry-run
|
||||
|
||||
**Result:** Safe preview of sync operations.
|
||||
|
||||
### Advanced: List Project Files by Route
|
||||
### Advanced: List Remote Files
|
||||
|
||||
**Use case:** Inspect local or cloud project files explicitly.
|
||||
**Use case:** See what files exist on cloud without syncing.
|
||||
|
||||
```bash
|
||||
# List local project files (default target when no route flag is given)
|
||||
# List all files in project
|
||||
bm project ls --name research
|
||||
bm project ls --name research --local
|
||||
|
||||
# List cloud project files
|
||||
bm project ls --name research --cloud
|
||||
|
||||
# List files in subdirectory
|
||||
bm project ls --name research --cloud --path subfolder
|
||||
bm project ls --name research --path subfolder
|
||||
```
|
||||
|
||||
**What happens:**
|
||||
1. Resolves route from flags (or local default when no route is given)
|
||||
2. Lists files for the chosen project instance
|
||||
1. Connects to cloud via rclone
|
||||
2. Lists files in remote project path
|
||||
3. No files transferred
|
||||
|
||||
**Result:** See file listing for the target route.
|
||||
**Result:** See cloud file listing.
|
||||
|
||||
## Multiple Projects
|
||||
|
||||
@@ -527,25 +390,25 @@ bm project ls --name research --cloud --path subfolder
|
||||
|
||||
```bash
|
||||
# Setup multiple projects
|
||||
bm project add research --cloud --local-path ~/Documents/research
|
||||
bm project add work --cloud --local-path ~/work-notes
|
||||
bm project add personal --cloud --local-path ~/personal
|
||||
bm project add research --local-path ~/Documents/research
|
||||
bm project add work --local-path ~/work-notes
|
||||
bm project add personal --local-path ~/personal
|
||||
|
||||
# Establish baselines
|
||||
bm cloud bisync --name research --resync
|
||||
bm cloud bisync --name work --resync
|
||||
bm cloud bisync --name personal --resync
|
||||
bm project bisync --name research --resync
|
||||
bm project bisync --name work --resync
|
||||
bm project bisync --name personal --resync
|
||||
|
||||
# Daily workflow: sync everything
|
||||
bm cloud bisync --name research
|
||||
bm cloud bisync --name work
|
||||
bm cloud bisync --name personal
|
||||
bm project bisync --name research
|
||||
bm project bisync --name work
|
||||
bm project bisync --name personal
|
||||
```
|
||||
|
||||
**Future:** `--all` flag will sync all configured projects:
|
||||
|
||||
```bash
|
||||
bm cloud bisync --all # Coming soon
|
||||
bm project bisync --all # Coming soon
|
||||
```
|
||||
|
||||
### Mixed Usage
|
||||
@@ -554,120 +417,36 @@ bm cloud bisync --all # Coming soon
|
||||
|
||||
```bash
|
||||
# Projects with sync
|
||||
bm project add research --cloud --local-path ~/Documents/research
|
||||
bm project add work --cloud --local-path ~/work
|
||||
bm project add research --local-path ~/Documents/research
|
||||
bm project add work --local-path ~/work
|
||||
|
||||
# Cloud-only projects
|
||||
bm project add archive --cloud
|
||||
bm project add temp-notes --cloud
|
||||
bm project add archive
|
||||
bm project add temp-notes
|
||||
|
||||
# Sync only the configured ones
|
||||
bm cloud bisync --name research
|
||||
bm cloud bisync --name work
|
||||
bm project bisync --name research
|
||||
bm project bisync --name work
|
||||
|
||||
# Archive and temp-notes stay cloud-only
|
||||
```
|
||||
|
||||
**Result:** Fine-grained control over what syncs.
|
||||
|
||||
## Per-Project Cloud Routing (API Key)
|
||||
## Disable Cloud Mode
|
||||
|
||||
Route individual projects through cloud using an API key. This lets you keep some projects local while others route through cloud.
|
||||
|
||||
### Setting Up API Key Auth
|
||||
|
||||
**Option A: Create a key in the web app, then save it locally:**
|
||||
|
||||
```bash
|
||||
bm cloud set-key bmc_abc123...
|
||||
```
|
||||
|
||||
**Option B: Create a key via CLI (requires OAuth login first):**
|
||||
|
||||
```bash
|
||||
bm cloud login # One-time OAuth login
|
||||
bm cloud create-key "my-laptop" # Creates key and saves it locally
|
||||
```
|
||||
|
||||
The API key is account-level — it grants access to all your cloud projects. It's stored in `~/.basic-memory/config.json` as `cloud_api_key`.
|
||||
On POSIX systems, Basic Memory writes `~/.basic-memory/` as user-private (`0700`) and
|
||||
`config.json` as user-read/write only (`0600`). Treat this config file as a credential
|
||||
file when an API key is saved.
|
||||
|
||||
### Setting Project Modes
|
||||
|
||||
```bash
|
||||
# Route a project through cloud
|
||||
bm project set-cloud research
|
||||
|
||||
# Revert to local mode
|
||||
bm project set-local research
|
||||
|
||||
# View project modes
|
||||
bm project list
|
||||
```
|
||||
|
||||
**What happens:**
|
||||
- `set-cloud`: validates the API key exists, then sets the project mode to `cloud` in config
|
||||
- `set-local`: reverts the project to local mode (removes the mode entry from config)
|
||||
- MCP tools and CLI commands for that project will route to `cloud_host/proxy` with the API key as Bearer token
|
||||
|
||||
### How It Works
|
||||
|
||||
When an MCP tool or CLI command runs for a cloud-mode project:
|
||||
|
||||
1. `get_client(project_name="research")` checks the project's mode in config
|
||||
2. If mode is `cloud`, creates an HTTP client pointed at `cloud_host/proxy` with `Authorization: Bearer bmc_...`
|
||||
3. If mode is `local` (default), uses the in-process ASGI transport as usual
|
||||
|
||||
**Routing priority** (highest to lowest):
|
||||
1. Factory injection (cloud app, tests)
|
||||
2. Explicit route override (`--local` / `--cloud`)
|
||||
3. Per-project cloud mode (API key)
|
||||
4. Local ASGI transport (default)
|
||||
|
||||
Route override environment variables:
|
||||
- `BASIC_MEMORY_FORCE_LOCAL=true`
|
||||
- `BASIC_MEMORY_FORCE_CLOUD=true`
|
||||
- `BASIC_MEMORY_EXPLICIT_ROUTING=true`
|
||||
|
||||
No-project, no-flag CLI commands default to local routing.
|
||||
|
||||
### Configuration Example
|
||||
|
||||
```json
|
||||
{
|
||||
"projects": {
|
||||
"personal": "/Users/me/notes",
|
||||
"research": "/Users/me/research"
|
||||
},
|
||||
"project_modes": {
|
||||
"research": "cloud"
|
||||
},
|
||||
"cloud_api_key": "bmc_abc123...",
|
||||
"cloud_host": "https://cloud.basicmemory.com",
|
||||
"default_project": "personal"
|
||||
}
|
||||
```
|
||||
|
||||
In this example, `personal` stays local and `research` routes through cloud. Projects not listed in `project_modes` default to local.
|
||||
|
||||
### Sync Behavior
|
||||
|
||||
Cloud-mode projects are automatically skipped during local file sync (background sync and file watching). Their files live on the cloud instance, not locally.
|
||||
|
||||
## OAuth Logout
|
||||
Return to local mode:
|
||||
|
||||
```bash
|
||||
bm cloud logout
|
||||
```
|
||||
|
||||
**What this does:**
|
||||
1. Removes stored OAuth token(s)
|
||||
2. Does not change per-project route configuration
|
||||
3. Does not change command routing defaults
|
||||
1. Disables cloud mode in config
|
||||
2. All commands now work locally
|
||||
3. Auth token remains (can re-enable with login)
|
||||
|
||||
**Result:** OAuth session is cleared. API-key-based routing still works if `cloud_api_key` is configured.
|
||||
**Result:** All `bm` commands work with local projects again.
|
||||
|
||||
## Filter Configuration
|
||||
|
||||
@@ -682,62 +461,33 @@ bm cloud logout
|
||||
**Default patterns:**
|
||||
|
||||
```gitignore
|
||||
# Hidden files and directories
|
||||
.*
|
||||
|
||||
# Basic Memory internals
|
||||
*.db
|
||||
*.db-shm
|
||||
*.db-wal
|
||||
config.json
|
||||
|
||||
# Version control
|
||||
.git
|
||||
.svn
|
||||
.git/**
|
||||
|
||||
# Python
|
||||
__pycache__
|
||||
__pycache__/**
|
||||
*.pyc
|
||||
*.pyo
|
||||
*.pyd
|
||||
.pytest_cache
|
||||
.coverage
|
||||
*.egg-info
|
||||
.tox
|
||||
.mypy_cache
|
||||
.ruff_cache
|
||||
|
||||
# Virtual environments
|
||||
.venv
|
||||
venv
|
||||
env
|
||||
.env
|
||||
.venv/**
|
||||
venv/**
|
||||
|
||||
# Node.js
|
||||
node_modules
|
||||
node_modules/**
|
||||
|
||||
# Build artifacts
|
||||
build
|
||||
dist
|
||||
.cache
|
||||
|
||||
# IDE
|
||||
.idea
|
||||
.vscode
|
||||
# Basic Memory internals
|
||||
memory.db/**
|
||||
memory.db-shm/**
|
||||
memory.db-wal/**
|
||||
config.json/**
|
||||
watch-status.json/**
|
||||
.bmignore.rclone/**
|
||||
|
||||
# OS files
|
||||
.DS_Store
|
||||
Thumbs.db
|
||||
desktop.ini
|
||||
.DS_Store/**
|
||||
Thumbs.db/**
|
||||
|
||||
# Obsidian
|
||||
.obsidian
|
||||
|
||||
# Temporary files
|
||||
*.tmp
|
||||
*.swp
|
||||
*.swo
|
||||
*~
|
||||
# Environment files
|
||||
.env/**
|
||||
.env.local/**
|
||||
```
|
||||
|
||||
**How it works:**
|
||||
@@ -746,11 +496,6 @@ desktop.ini
|
||||
3. Rclone uses filters during sync
|
||||
4. Same patterns used by all projects
|
||||
|
||||
During conversion, file patterns exclude the direct match and recursive contents.
|
||||
For example, `config.json` becomes both `- config.json` and `- config.json/**`,
|
||||
while `.*` becomes both `- .*` and `- .*/**`. Directory-only patterns keep
|
||||
their trailing slash, so `cache/` becomes `- cache/` and `- cache/**`.
|
||||
|
||||
**Customizing:**
|
||||
|
||||
```bash
|
||||
@@ -758,41 +503,14 @@ their trailing slash, so `cache/` becomes `- cache/` and `- cache/**`.
|
||||
code ~/.basic-memory/.bmignore
|
||||
|
||||
# Add custom patterns
|
||||
echo "*.tmp" >> ~/.basic-memory/.bmignore
|
||||
echo "*.tmp/**" >> ~/.basic-memory/.bmignore
|
||||
|
||||
# Next sync uses updated patterns
|
||||
bm cloud bisync --name research
|
||||
bm project bisync --name research
|
||||
```
|
||||
|
||||
## Troubleshooting
|
||||
|
||||
### Rclone Setup Cannot Install Automatically
|
||||
|
||||
**Problem:** `bm cloud setup` cannot find a supported package manager, or package-manager
|
||||
installation fails.
|
||||
|
||||
**Explanation:** The CLI avoids remote privileged install scripts. It only invokes known
|
||||
package managers and otherwise asks you to install rclone manually.
|
||||
|
||||
**Solution:** Install rclone with your OS package manager, then rerun setup:
|
||||
|
||||
```bash
|
||||
# macOS
|
||||
brew install rclone
|
||||
|
||||
# Debian/Ubuntu
|
||||
sudo apt install rclone
|
||||
|
||||
# Fedora
|
||||
sudo dnf install rclone
|
||||
|
||||
# Arch
|
||||
sudo pacman -S rclone
|
||||
|
||||
# After rclone is on PATH
|
||||
bm cloud setup
|
||||
```
|
||||
|
||||
### Authentication Issues
|
||||
|
||||
**Problem:** "Authentication failed" or "Invalid token"
|
||||
@@ -824,7 +542,7 @@ bm cloud login
|
||||
**Solution:**
|
||||
|
||||
```bash
|
||||
bm cloud bisync --name research --resync
|
||||
bm project bisync --name research --resync
|
||||
```
|
||||
|
||||
**What this does:**
|
||||
@@ -847,7 +565,7 @@ bm cloud bisync --name research --resync
|
||||
echo "# Research Notes" > ~/Documents/research/README.md
|
||||
|
||||
# Now run bisync
|
||||
bm cloud bisync --name research --resync
|
||||
bm project bisync --name research --resync
|
||||
```
|
||||
|
||||
**Why this happens:** Bisync creates listing files that track the state of each side. When both directories are completely empty, these listing files are considered invalid by rclone.
|
||||
@@ -864,10 +582,10 @@ bm cloud bisync --name research --resync
|
||||
|
||||
```bash
|
||||
# Clear bisync state
|
||||
bm cloud bisync-reset research
|
||||
bm project bisync-reset research
|
||||
|
||||
# Re-establish baseline
|
||||
bm cloud bisync --name research --resync
|
||||
bm project bisync --name research --resync
|
||||
```
|
||||
|
||||
**What this does:**
|
||||
@@ -887,16 +605,16 @@ bm cloud bisync --name research --resync
|
||||
|
||||
```bash
|
||||
# Check what would be deleted
|
||||
bm cloud bisync --name research --dry-run
|
||||
bm project bisync --name research --dry-run
|
||||
|
||||
# If correct, establish new baseline
|
||||
bm cloud bisync --name research --resync
|
||||
bm project bisync --name research --resync
|
||||
```
|
||||
|
||||
**Solution 2:** Use one-way sync if you know local is correct:
|
||||
|
||||
```bash
|
||||
bm cloud sync --name research
|
||||
bm project sync --name research
|
||||
```
|
||||
|
||||
### Project Not Configured for Sync
|
||||
@@ -908,8 +626,8 @@ bm cloud sync --name research
|
||||
**Solution:**
|
||||
|
||||
```bash
|
||||
bm cloud sync-setup research ~/Documents/research
|
||||
bm cloud bisync --name research --resync
|
||||
bm project sync-setup research ~/Documents/research
|
||||
bm project bisync --name research --resync
|
||||
```
|
||||
|
||||
### Connection Issues
|
||||
@@ -928,116 +646,78 @@ If instance is down, wait a few minutes and retry.
|
||||
|
||||
- **Authentication**: OAuth 2.1 with PKCE flow
|
||||
- **Tokens**: Stored securely in `~/.basic-memory/basic-memory-cloud.json`
|
||||
- **API keys**: Stored in `~/.basic-memory/config.json`, which is written with private file permissions on POSIX systems
|
||||
- **Transport**: All data encrypted in transit (HTTPS)
|
||||
- **Credentials**: Scoped S3 credentials (read-write to your tenant only)
|
||||
- **Rclone setup**: Uses package managers or manual instructions; no remote privileged install-script fallback
|
||||
- **Isolation**: Your data isolated from other tenants
|
||||
- **Ignore patterns**: Sensitive files automatically excluded via `.bmignore`
|
||||
|
||||
## Command Reference
|
||||
|
||||
### Cloud Authentication
|
||||
### Cloud Mode Management
|
||||
|
||||
```bash
|
||||
bm cloud login # Authenticate and store OAuth credentials
|
||||
bm cloud logout # Remove stored OAuth credentials
|
||||
bm cloud status # Check auth state and instance health
|
||||
bm cloud promo --off # Disable CLI cloud promo notices
|
||||
```
|
||||
|
||||
### API Key Management
|
||||
|
||||
```bash
|
||||
bm cloud set-key <key> # Save a cloud API key (bmc_ prefixed)
|
||||
bm cloud create-key <name> # Create API key via cloud API (requires OAuth login)
|
||||
bm cloud login # Authenticate and enable cloud mode
|
||||
bm cloud logout # Disable cloud mode
|
||||
bm cloud status # Check cloud mode and instance health
|
||||
```
|
||||
|
||||
### Setup
|
||||
|
||||
```bash
|
||||
bm cloud setup # Install rclone via package manager and configure credentials
|
||||
bm cloud setup # Install rclone and configure credentials
|
||||
```
|
||||
|
||||
### Project Management
|
||||
|
||||
When cloud mode is enabled:
|
||||
|
||||
```bash
|
||||
bm project list --local # Local project list
|
||||
bm project list --cloud # Cloud project list
|
||||
bm project add <name> --cloud # Create cloud project (no sync)
|
||||
bm project add <name> --cloud --local-path <path> # Create with local sync
|
||||
bm cloud sync-setup <name> <path> # Add sync to existing project
|
||||
bm project list # List cloud projects
|
||||
bm project add <name> # Create cloud project (no sync)
|
||||
bm project add <name> --local-path <path> # Create with local sync
|
||||
bm project sync-setup <name> <path> # Add sync to existing project
|
||||
bm project rm <name> # Delete project
|
||||
```
|
||||
|
||||
### Per-Project Routing
|
||||
|
||||
```bash
|
||||
bm project set-cloud <name> # Route project through cloud (requires API key)
|
||||
bm project set-local <name> # Revert project to local mode
|
||||
```
|
||||
|
||||
### File Synchronization
|
||||
|
||||
```bash
|
||||
# Pull: fetch cloud changes (cloud → local) - Personal + Team, additive
|
||||
bm cloud pull --name <project>
|
||||
bm cloud pull --name <project> --dry-run
|
||||
bm cloud pull --name <project> --on-conflict [fail|keep-local|keep-cloud|keep-both]
|
||||
# One-way sync (local → cloud)
|
||||
bm project sync --name <project>
|
||||
bm project sync --name <project> --dry-run
|
||||
bm project sync --name <project> --verbose
|
||||
|
||||
# Push: upload local changes (local → cloud) - Personal + Team, additive
|
||||
bm cloud push --name <project>
|
||||
bm cloud push --name <project> --dry-run
|
||||
bm cloud push --name <project> --on-conflict [fail|keep-local|keep-cloud|keep-both]
|
||||
# Two-way sync (local ↔ cloud) - Recommended
|
||||
bm project bisync --name <project> # After first --resync
|
||||
bm project bisync --name <project> --resync # First time / force baseline
|
||||
bm project bisync --name <project> --dry-run
|
||||
bm project bisync --name <project> --verbose
|
||||
|
||||
# One-way mirror (local → cloud) - Personal workspaces only
|
||||
bm cloud sync --name <project>
|
||||
bm cloud sync --name <project> --dry-run
|
||||
bm cloud sync --name <project> --verbose
|
||||
# Integrity check
|
||||
bm project check --name <project>
|
||||
bm project check --name <project> --one-way
|
||||
|
||||
# Two-way mirror (local ↔ cloud) - Personal workspaces only
|
||||
bm cloud bisync --name <project> # After first --resync
|
||||
bm cloud bisync --name <project> --resync # First time / force baseline
|
||||
bm cloud bisync --name <project> --dry-run
|
||||
bm cloud bisync --name <project> --verbose
|
||||
|
||||
# Integrity check - Personal workspaces only
|
||||
bm cloud check --name <project>
|
||||
bm cloud check --name <project> --one-way
|
||||
|
||||
# List project files by route
|
||||
bm project ls --name <project> # Default target: local
|
||||
bm project ls --name <project> --local
|
||||
bm project ls --name <project> --cloud
|
||||
bm project ls --name <project> --cloud --path <subpath>
|
||||
# List remote files
|
||||
bm project ls --name <project>
|
||||
bm project ls --name <project> --path <subpath>
|
||||
```
|
||||
|
||||
## Summary
|
||||
|
||||
**Basic Memory Cloud uses project-scoped sync:**
|
||||
|
||||
1. **Authenticate cloud access** - `bm cloud login`
|
||||
1. **Enable cloud mode** - `bm cloud login`
|
||||
2. **Install rclone** - `bm cloud setup`
|
||||
3. **Add projects with sync** - `bm project add research --cloud --local-path ~/Documents/research`
|
||||
|
||||
**Personal workspace (solo, mirror) workflow:**
|
||||
|
||||
4. **Preview first sync** - `bm cloud bisync --name research --resync --dry-run`
|
||||
5. **Establish baseline** - `bm cloud bisync --name research --resync`
|
||||
6. **Daily workflow** - `bm cloud bisync --name research`
|
||||
|
||||
**Team workspace (shared, additive) workflow:**
|
||||
|
||||
4. **Fetch teammates' changes** - `bm cloud pull --name research`
|
||||
5. **Upload your changes** - `bm cloud push --name research`
|
||||
6. **Resolve conflicts explicitly** - re-run with `--on-conflict keep-cloud|keep-local|keep-both`
|
||||
3. **Add projects with sync** - `bm project add research --local-path ~/Documents/research`
|
||||
4. **Preview first sync** - `bm project bisync --name research --resync --dry-run`
|
||||
5. **Establish baseline** - `bm project bisync --name research --resync`
|
||||
6. **Daily workflow** - `bm project bisync --name research`
|
||||
|
||||
**Key benefits:**
|
||||
- ✅ Each project independently syncs (or doesn't)
|
||||
- ✅ Projects can live anywhere on disk
|
||||
- ✅ Explicit sync operations (no magic)
|
||||
- ✅ Team-safe push/pull that never delete on the destination
|
||||
- ✅ Safe by design (max delete limits, conflict resolution, git-style conflict aborts)
|
||||
- ✅ Safe by design (max delete limits, conflict resolution)
|
||||
- ✅ Full offline access (work locally, sync when ready)
|
||||
|
||||
**Future enhancements:**
|
||||
|
||||
@@ -1,91 +0,0 @@
|
||||
# Cloud Semantic Search Value (Customer-Facing Technical Story)
|
||||
|
||||
This document explains why teams should buy cloud semantic search even when local search exists.
|
||||
|
||||
## Core Promise
|
||||
|
||||
Markdown files remain the source of truth in both local and cloud modes.
|
||||
|
||||
- Files are portable.
|
||||
- Search indexes are derived and rebuildable.
|
||||
- You never get locked into proprietary document storage.
|
||||
|
||||
## The Customer Problem
|
||||
|
||||
Teams paying for cloud are usually not optimizing for "can this run locally." They are optimizing for:
|
||||
|
||||
- finding the right note the first time,
|
||||
- keeping retrieval quality high as note volume grows,
|
||||
- avoiding search slowdowns while content is actively changing,
|
||||
- getting consistent results across users, agents, and sessions.
|
||||
|
||||
## Why Cloud Is the Aspirin
|
||||
|
||||
Cloud semantic search is the immediate pain reliever because it fixes the problems users feel right now.
|
||||
|
||||
### 1) Better hit rate on real queries
|
||||
|
||||
Cloud uses stronger managed embeddings than the default local model, which improves semantic recall for paraphrases and vague questions.
|
||||
|
||||
Customer outcome:
|
||||
|
||||
- fewer "I know this exists but search missed it" moments,
|
||||
- less query rewording,
|
||||
- faster time to answer.
|
||||
|
||||
### 2) Better behavior under active workloads
|
||||
|
||||
Cloud indexing runs out of band in workers, so indexing does not compete with interactive read/write traffic.
|
||||
|
||||
Customer outcome:
|
||||
|
||||
- stable search responsiveness during heavy updates,
|
||||
- fresher semantic results shortly after edits,
|
||||
- less user-visible performance variance.
|
||||
|
||||
### 3) Better consistency for shared knowledge
|
||||
|
||||
Cloud retrieval runs against a centralized tenant index, so teams and agents resolve against the same semantic state.
|
||||
|
||||
Customer outcome:
|
||||
|
||||
- fewer "works on my machine" search differences,
|
||||
- more predictable agent behavior across environments,
|
||||
- easier cross-user collaboration on large knowledge bases.
|
||||
|
||||
### 4) Better quality at higher scale
|
||||
|
||||
With Postgres + `pgvector` per tenant, cloud can sustain larger note collections and higher query volumes than typical local setups.
|
||||
|
||||
Customer outcome:
|
||||
|
||||
- confidence as repositories grow to tens of thousands of notes,
|
||||
- less need for user-side tuning,
|
||||
- fewer quality regressions as usage increases.
|
||||
|
||||
## Local Is the Vitamin
|
||||
|
||||
Local semantic search still matters and should stay strong.
|
||||
|
||||
- offline use,
|
||||
- privacy-first operation,
|
||||
- no cloud dependency,
|
||||
- user-controlled runtime.
|
||||
|
||||
It compounds long-term ownership and resilience, but does not remove the immediate pain points cloud solves for teams at scale.
|
||||
|
||||
## Recommended Messaging
|
||||
|
||||
One-liner:
|
||||
|
||||
"Cloud semantic search is the aspirin: it fixes retrieval quality and performance pain now. Local semantic search is the vitamin: it builds long-term control and resilience."
|
||||
|
||||
Long form:
|
||||
|
||||
"Basic Memory keeps markdown as the source of truth everywhere. Local gives privacy and offline control. Cloud adds immediate, measurable improvements in search quality, consistency, and responsiveness for teams and agents running at scale."
|
||||
|
||||
## Packaging Guidance
|
||||
|
||||
- Base: local FTS plus optional local semantic search.
|
||||
- Cloud value: higher semantic quality, stable performance under load, and consistent team-wide retrieval.
|
||||
- Keep interfaces pluggable (`EmbeddingProvider`, vector backend protocol) so implementation can evolve without changing user workflows.
|
||||
@@ -1,300 +0,0 @@
|
||||
# LiteLLM Provider
|
||||
|
||||
Basic Memory can use the LiteLLM SDK for semantic search embeddings. This lets you
|
||||
keep Basic Memory's vector indexing and search behavior while routing embedding calls
|
||||
to OpenAI-compatible and provider-specific backends such as OpenAI, Azure OpenAI,
|
||||
Cohere, Bedrock, NVIDIA NIM, and other LiteLLM-supported embedding providers.
|
||||
|
||||
Use this page when you want to try a non-default embedding model, validate a provider,
|
||||
or tune LiteLLM-specific settings.
|
||||
|
||||
> **Experimental — advanced users only.** The LiteLLM provider is experimental and
|
||||
> intended for users who are comfortable operating remote embedding backends. It makes
|
||||
> paid, networked API calls, requires per-model dimension and input-role configuration,
|
||||
> and reindexing a real corpus can be slow and spend provider quota (see
|
||||
> [Reindexing with a remote provider](#reindexing-with-a-remote-provider)). For most
|
||||
> users, the default local **FastEmbed** provider is the recommended choice. Use LiteLLM
|
||||
> only if you know what you're doing.
|
||||
|
||||
## Quick Start
|
||||
|
||||
The default LiteLLM model is OpenAI `text-embedding-3-small` through the LiteLLM
|
||||
model string `openai/text-embedding-3-small`.
|
||||
|
||||
```bash
|
||||
export BASIC_MEMORY_SEMANTIC_SEARCH_ENABLED=true
|
||||
export BASIC_MEMORY_SEMANTIC_EMBEDDING_PROVIDER=litellm
|
||||
export OPENAI_API_KEY=sk-...
|
||||
|
||||
bm reindex --embeddings
|
||||
```
|
||||
|
||||
Then use vector or hybrid search:
|
||||
|
||||
```python
|
||||
search_notes("login token flow", search_type="hybrid")
|
||||
```
|
||||
|
||||
## Basic Memory Options
|
||||
|
||||
All options can be set in config or as environment variables.
|
||||
|
||||
| Config Field | Env Var | Default | Notes |
|
||||
|---|---|---|---|
|
||||
| `semantic_search_enabled` | `BASIC_MEMORY_SEMANTIC_SEARCH_ENABLED` | Auto | Set to `true` to force vector/hybrid support on. |
|
||||
| `semantic_embedding_provider` | `BASIC_MEMORY_SEMANTIC_EMBEDDING_PROVIDER` | `fastembed` | Set to `litellm` for the LiteLLM provider. |
|
||||
| `semantic_embedding_model` | `BASIC_MEMORY_SEMANTIC_EMBEDDING_MODEL` | `bge-small-en-v1.5` | With `litellm`, the default is remapped to `openai/text-embedding-3-small`. |
|
||||
| `semantic_embedding_dimensions` | `BASIC_MEMORY_SEMANTIC_EMBEDDING_DIMENSIONS` | Provider default | Required for non-default LiteLLM models because vector tables are dimensioned before the first API call. |
|
||||
| `semantic_embedding_forward_dimensions` | `BASIC_MEMORY_SEMANTIC_EMBEDDING_FORWARD_DIMENSIONS` | Auto | Sends `dimensions` to LiteLLM only when supported. Auto is enabled for `text-embedding-3` model strings. |
|
||||
| `semantic_embedding_document_input_type` | `BASIC_MEMORY_SEMANTIC_EMBEDDING_DOCUMENT_INPUT_TYPE` | Auto | LiteLLM `input_type` for indexed notes/passages. |
|
||||
| `semantic_embedding_query_input_type` | `BASIC_MEMORY_SEMANTIC_EMBEDDING_QUERY_INPUT_TYPE` | Auto | LiteLLM `input_type` for search queries. |
|
||||
| `semantic_embedding_batch_size` | `BASIC_MEMORY_SEMANTIC_EMBEDDING_BATCH_SIZE` | `2` | Number of text chunks per provider request. |
|
||||
| `semantic_embedding_request_concurrency` | `BASIC_MEMORY_SEMANTIC_EMBEDDING_REQUEST_CONCURRENCY` | `4` | Maximum concurrent LiteLLM embedding requests. |
|
||||
| `semantic_embedding_sync_batch_size` | `BASIC_MEMORY_SEMANTIC_EMBEDDING_SYNC_BATCH_SIZE` | `2` | Number of prepared vector jobs flushed through the sync pipeline together. |
|
||||
|
||||
## Dimensions
|
||||
|
||||
Basic Memory needs the vector dimension before it can create SQLite or Postgres
|
||||
vector tables. The OpenAI default is known, so this works without an explicit
|
||||
dimension:
|
||||
|
||||
```bash
|
||||
export BASIC_MEMORY_SEMANTIC_EMBEDDING_PROVIDER=litellm
|
||||
export BASIC_MEMORY_SEMANTIC_EMBEDDING_MODEL=openai/text-embedding-3-small
|
||||
```
|
||||
|
||||
For every other LiteLLM model, set the dimension explicitly:
|
||||
|
||||
```bash
|
||||
export BASIC_MEMORY_SEMANTIC_EMBEDDING_PROVIDER=litellm
|
||||
export BASIC_MEMORY_SEMANTIC_EMBEDDING_MODEL=cohere/embed-english-v3.0
|
||||
export BASIC_MEMORY_SEMANTIC_EMBEDDING_DIMENSIONS=1024
|
||||
```
|
||||
|
||||
For fixed-size models, `semantic_embedding_dimensions` is Basic Memory's local
|
||||
schema and validation size. For OpenAI/Azure `text-embedding-3` models, LiteLLM
|
||||
can also forward `dimensions` as a provider-side reduced-output request. Basic
|
||||
Memory enables that automatically when the model string contains `text-embedding-3`.
|
||||
|
||||
If you use an Azure deployment alias such as `azure/<deployment-name>`, the model
|
||||
string may not reveal that the underlying model supports reduced output dimensions.
|
||||
Set this only when your deployment supports it:
|
||||
|
||||
```bash
|
||||
export BASIC_MEMORY_SEMANTIC_EMBEDDING_FORWARD_DIMENSIONS=true
|
||||
```
|
||||
|
||||
## Asymmetric Models
|
||||
|
||||
Some embedding models use different request roles for indexed documents and
|
||||
search queries. Basic Memory automatically sets these for known LiteLLM families:
|
||||
|
||||
| Model Family | Document `input_type` | Query `input_type` |
|
||||
|---|---|---|
|
||||
| Cohere v3 embeddings | `search_document` | `search_query` |
|
||||
| NVIDIA NIM retrieval embeddings | `passage` | `query` |
|
||||
|
||||
For any other asymmetric model, configure both roles explicitly:
|
||||
|
||||
```bash
|
||||
export BASIC_MEMORY_SEMANTIC_EMBEDDING_DOCUMENT_INPUT_TYPE=passage
|
||||
export BASIC_MEMORY_SEMANTIC_EMBEDDING_QUERY_INPUT_TYPE=query
|
||||
```
|
||||
|
||||
Changing provider, model, dimensions, dimension-forwarding, or document/query
|
||||
roles changes the meaning of stored vectors. Rebuild embeddings after any of
|
||||
those changes:
|
||||
|
||||
```bash
|
||||
bm reindex --embeddings
|
||||
```
|
||||
|
||||
## Reindexing with a remote provider
|
||||
|
||||
Embedding a real corpus through a network API is far slower than local FastEmbed, and
|
||||
the defaults are tuned for the local case. Two things to know before you run a full
|
||||
reindex.
|
||||
|
||||
**Raise the sync batch size.** `semantic_embedding_sync_batch_size` defaults to `2`, and
|
||||
it — not `semantic_embedding_batch_size` — governs throughput on the sync pipeline. With
|
||||
the default, a full reindex can take tens of seconds *per note* against a remote provider.
|
||||
Raising both to a larger value turns a multi-minute (or longer) reindex into well under a
|
||||
minute for the same corpus:
|
||||
|
||||
```bash
|
||||
export BASIC_MEMORY_SEMANTIC_EMBEDDING_SYNC_BATCH_SIZE=32
|
||||
export BASIC_MEMORY_SEMANTIC_EMBEDDING_BATCH_SIZE=64
|
||||
```
|
||||
|
||||
Stay within the provider's per-request size and rate limits — Cohere v3, for example,
|
||||
accepts up to 96 inputs per embedding request.
|
||||
|
||||
**Changing dimensions requires recreating the vector table.** Basic Memory dimensions the
|
||||
vector table on first index and refuses to mix sizes. Switching to a model with a
|
||||
different dimension (for example FastEmbed 384 → OpenAI 1536 → Cohere 1024) makes a plain
|
||||
`bm reindex` raise an `Embedding dimension mismatch` error. Recreate the table with a full
|
||||
rebuild — files are the source of truth, so this re-indexes from disk and re-embeds
|
||||
everything:
|
||||
|
||||
```bash
|
||||
bm reset --reindex
|
||||
```
|
||||
|
||||
To trial a provider without disturbing your existing index, point Basic Memory at a
|
||||
throwaway config + database instead:
|
||||
|
||||
```bash
|
||||
export BASIC_MEMORY_CONFIG_DIR=/tmp/bm-litellm-trial
|
||||
```
|
||||
|
||||
## Provider Setup Examples
|
||||
|
||||
LiteLLM reads provider credentials from the environment. These are the examples
|
||||
covered by Basic Memory's live validation harness.
|
||||
|
||||
### OpenAI Through LiteLLM
|
||||
|
||||
```bash
|
||||
export OPENAI_API_KEY=sk-...
|
||||
export BASIC_MEMORY_SEMANTIC_SEARCH_ENABLED=true
|
||||
export BASIC_MEMORY_SEMANTIC_EMBEDDING_PROVIDER=litellm
|
||||
export BASIC_MEMORY_SEMANTIC_EMBEDDING_MODEL=openai/text-embedding-3-small
|
||||
```
|
||||
|
||||
### Cohere v3
|
||||
|
||||
```bash
|
||||
export COHERE_API_KEY=...
|
||||
export BASIC_MEMORY_SEMANTIC_SEARCH_ENABLED=true
|
||||
export BASIC_MEMORY_SEMANTIC_EMBEDDING_PROVIDER=litellm
|
||||
export BASIC_MEMORY_SEMANTIC_EMBEDDING_MODEL=cohere/embed-english-v3.0
|
||||
export BASIC_MEMORY_SEMANTIC_EMBEDDING_DIMENSIONS=1024
|
||||
```
|
||||
|
||||
The provider auto-selects `search_document` for indexed chunks and `search_query`
|
||||
for search queries.
|
||||
|
||||
### Azure OpenAI
|
||||
|
||||
```bash
|
||||
export AZURE_API_KEY=...
|
||||
export AZURE_API_BASE=https://<resource-name>.openai.azure.com
|
||||
export AZURE_API_VERSION=2024-02-01
|
||||
|
||||
export BASIC_MEMORY_SEMANTIC_SEARCH_ENABLED=true
|
||||
export BASIC_MEMORY_SEMANTIC_EMBEDDING_PROVIDER=litellm
|
||||
export BASIC_MEMORY_SEMANTIC_EMBEDDING_MODEL=azure/<deployment-name>
|
||||
export BASIC_MEMORY_SEMANTIC_EMBEDDING_DIMENSIONS=1536
|
||||
```
|
||||
|
||||
If your Azure deployment is a reduced-dimension `text-embedding-3` deployment,
|
||||
set the dimension you want and enable forwarding:
|
||||
|
||||
```bash
|
||||
export BASIC_MEMORY_SEMANTIC_EMBEDDING_DIMENSIONS=512
|
||||
export BASIC_MEMORY_SEMANTIC_EMBEDDING_FORWARD_DIMENSIONS=true
|
||||
```
|
||||
|
||||
### NVIDIA NIM
|
||||
|
||||
```bash
|
||||
export NVIDIA_NIM_API_KEY=...
|
||||
# Optional when using a custom or self-hosted NIM endpoint:
|
||||
export NVIDIA_NIM_API_BASE=https://integrate.api.nvidia.com/v1
|
||||
|
||||
export BASIC_MEMORY_SEMANTIC_SEARCH_ENABLED=true
|
||||
export BASIC_MEMORY_SEMANTIC_EMBEDDING_PROVIDER=litellm
|
||||
export BASIC_MEMORY_SEMANTIC_EMBEDDING_MODEL=nvidia_nim/nvidia/embed-qa-4
|
||||
export BASIC_MEMORY_SEMANTIC_EMBEDDING_DIMENSIONS=1024
|
||||
```
|
||||
|
||||
The provider auto-selects `passage` for indexed chunks and `query` for search
|
||||
queries.
|
||||
|
||||
## Testing LiteLLM Providers
|
||||
|
||||
Run the non-live LiteLLM unit and harness tests first:
|
||||
|
||||
```bash
|
||||
uv run pytest tests/repository/test_litellm_provider.py \
|
||||
test-int/semantic/test_litellm_live_harness.py -q
|
||||
```
|
||||
|
||||
Run the SQLite and Postgres vector identity regressions when changing model
|
||||
identity, role, or vector sync behavior:
|
||||
|
||||
```bash
|
||||
uv run pytest \
|
||||
tests/repository/test_sqlite_vector_search_repository.py::test_sqlite_embedding_model_key_includes_litellm_role_settings \
|
||||
-q
|
||||
|
||||
BASIC_MEMORY_TEST_POSTGRES=1 uv run pytest \
|
||||
tests/repository/test_postgres_search_repository.py::test_postgres_litellm_role_change_reembeds_existing_chunks \
|
||||
-q
|
||||
```
|
||||
|
||||
The Postgres command uses testcontainers, so Docker must be running.
|
||||
|
||||
## Live Provider Harness
|
||||
|
||||
The live harness makes real LiteLLM API calls and spends provider quota. It is
|
||||
opt-in by design:
|
||||
|
||||
```bash
|
||||
export OPENAI_API_KEY=sk-...
|
||||
export COHERE_API_KEY=...
|
||||
|
||||
just test-litellm-live
|
||||
```
|
||||
|
||||
Built-in cases run when their API keys are present:
|
||||
|
||||
| Case | Required Env Var | Validates |
|
||||
|---|---|---|
|
||||
| `openai-text-embedding-3-small` | `OPENAI_API_KEY` | OpenAI via LiteLLM, 1536 dimensions, normalized vectors, ranking sanity. |
|
||||
| `cohere-embed-english-v3` | `COHERE_API_KEY` | Cohere v3 role handling, 1024 dimensions, normalized vectors, ranking sanity. |
|
||||
|
||||
Add provider aliases or new backends with a custom cases file:
|
||||
|
||||
```bash
|
||||
cat > /tmp/litellm-cases.json <<'JSON'
|
||||
[
|
||||
{
|
||||
"name": "azure-text-embedding-3-small-512",
|
||||
"model": "azure/<deployment-name>",
|
||||
"dimensions": 512,
|
||||
"api_key_env": "AZURE_API_KEY",
|
||||
"forward_dimensions": true
|
||||
},
|
||||
{
|
||||
"name": "nvidia-embed-qa-4",
|
||||
"model": "nvidia_nim/nvidia/embed-qa-4",
|
||||
"dimensions": 1024,
|
||||
"api_key_env": "NVIDIA_NIM_API_KEY",
|
||||
"document_input_type": "passage",
|
||||
"query_input_type": "query"
|
||||
}
|
||||
]
|
||||
JSON
|
||||
|
||||
just test-litellm-live --cases-file /tmp/litellm-cases.json
|
||||
```
|
||||
|
||||
For CI-style output:
|
||||
|
||||
```bash
|
||||
just test-litellm-live --cases-file /tmp/litellm-cases.json --json
|
||||
```
|
||||
|
||||
The harness embeds two documents and one query, validates dimension and vector
|
||||
normalization, checks that the authentication query ranks the authentication
|
||||
document above a distractor, and reports latency plus role/dimension settings.
|
||||
|
||||
## Provider Reference
|
||||
|
||||
LiteLLM's own provider and embedding docs are the source of truth for current
|
||||
model strings and credential names:
|
||||
|
||||
- [LiteLLM embedding models](https://docs.litellm.ai/docs/embedding/supported_embedding)
|
||||
- [LiteLLM Azure OpenAI provider](https://docs.litellm.ai/docs/providers/azure)
|
||||
- [LiteLLM NVIDIA NIM provider](https://docs.litellm.ai/docs/providers/nvidia_nim)
|
||||
@@ -1,499 +0,0 @@
|
||||
# Logfire Instrumentation Strategy
|
||||
|
||||
## Why
|
||||
|
||||
We want Logfire in Basic Memory for two specific use cases:
|
||||
|
||||
1. Local development and performance investigation
|
||||
2. Cloud deployments where Basic Memory runs inside Basic Memory Cloud
|
||||
|
||||
This instrumentation must be:
|
||||
|
||||
- Disabled by default
|
||||
- Useful when enabled
|
||||
- Safe for local-first users
|
||||
- Searchable in Logfire over time
|
||||
|
||||
The previous integration added telemetry, but it leaned too much on generic framework instrumentation. That created noisy spans with weak names and made the trace view harder to navigate. This strategy favors manual instrumentation around Basic Memory's real units of work.
|
||||
|
||||
## Core Principles
|
||||
|
||||
### 1. Default-off
|
||||
|
||||
Basic Memory should ship with Logfire disabled unless the operator explicitly enables it.
|
||||
|
||||
That means:
|
||||
|
||||
- no required token for normal local usage
|
||||
- no surprise outbound telemetry
|
||||
- no behavior change for existing users
|
||||
|
||||
### 2. Manual spans over automatic framework spans
|
||||
|
||||
We should not rely on broad auto-instrumentation for FastAPI, MCP, SQLAlchemy, or HTTP as the primary experience.
|
||||
|
||||
Why:
|
||||
|
||||
- auto-generated span names are often generic
|
||||
- routes and middleware produce too many low-signal spans
|
||||
- it becomes harder to answer product questions like "why was `write_note` slow?" or "where did sync time go?"
|
||||
|
||||
The preferred model is:
|
||||
|
||||
- one meaningful root span per high-level operation
|
||||
- a small number of child spans for important phases
|
||||
- optional targeted instrumentation only where it adds clear value
|
||||
|
||||
### 3. Logs must live inside traces
|
||||
|
||||
Basic Memory already uses `loguru` pervasively. The Logfire integration should preserve that and make those logs visible inside the active trace/span context.
|
||||
|
||||
If traces exist but the logs are detached from them, the integration is not doing its job.
|
||||
|
||||
### 4. Stable names, selective attributes
|
||||
|
||||
Span names should describe the operation class, not the specific input.
|
||||
|
||||
Good:
|
||||
|
||||
- `mcp.tool.write_note`
|
||||
- `sync.project.scan`
|
||||
- `search.execute`
|
||||
- `routing.resolve_project`
|
||||
|
||||
Bad:
|
||||
|
||||
- `Searching for "foo bar baz"`
|
||||
- `POST /v2/projects/123/search/`
|
||||
- `write note to /specs/api.md`
|
||||
|
||||
Dynamic values belong in attributes, not in the span name.
|
||||
|
||||
## What We Should Not Do
|
||||
|
||||
### Avoid broad FastAPI auto-instrumentation
|
||||
|
||||
We should not turn on `instrument_fastapi()` and treat that as the main telemetry story.
|
||||
|
||||
It may still be useful in narrowly scoped debugging, but it should not define the production trace shape. The meaningful root spans should come from Basic Memory's own entrypoints and service boundaries.
|
||||
|
||||
### Avoid per-file spans by default
|
||||
|
||||
`sync` can process many files. A span per file will explode trace cardinality and make performance views noisy.
|
||||
|
||||
Default behavior should be:
|
||||
|
||||
- one span for the project sync
|
||||
- child spans for scan, move handling, delete handling, markdown sync batch, relation resolution, embedding sync, watermark update
|
||||
- per-file spans only for failures or very slow outliers
|
||||
|
||||
### Avoid high-cardinality attributes on every span
|
||||
|
||||
Do not attach large or highly variable values everywhere:
|
||||
|
||||
- raw note content
|
||||
- file bodies
|
||||
- long search text
|
||||
- arbitrary metadata blobs
|
||||
- unique IDs that make every span shape distinct
|
||||
|
||||
Prefer compact, queryable attributes:
|
||||
|
||||
- `project_name`
|
||||
- `workspace_id`
|
||||
- `route_mode`
|
||||
- `scan_type`
|
||||
- `file_count`
|
||||
- `result_count`
|
||||
- `search_type`
|
||||
- `retrieval_mode`
|
||||
- `duration_ms`
|
||||
|
||||
## Proposed Architecture
|
||||
|
||||
Add a dedicated telemetry module in core Basic Memory, separate from logging setup.
|
||||
|
||||
Suggested shape:
|
||||
|
||||
```python
|
||||
# basic_memory/telemetry.py
|
||||
|
||||
def configure_telemetry(service_name: str, *, enable_logfire: bool) -> None: ...
|
||||
def telemetry_enabled() -> bool: ...
|
||||
def span(name: str, **attrs): ...
|
||||
def bind_telemetry_context(**attrs): ...
|
||||
```
|
||||
|
||||
This module should:
|
||||
|
||||
- configure Logfire only when explicitly enabled
|
||||
- set up the Logfire `loguru` handler
|
||||
- expose lightweight helpers so application code does not import `logfire` directly everywhere
|
||||
- degrade cleanly to no-op behavior when disabled
|
||||
|
||||
This keeps the rest of the codebase readable and makes it easy to reason about what telemetry is doing.
|
||||
|
||||
## Logging Integration Strategy
|
||||
|
||||
### Goal
|
||||
|
||||
When a span is active, logs emitted through `loguru` during that operation should show up in the same trace.
|
||||
|
||||
### Preferred design
|
||||
|
||||
1. Configure Logfire once in the telemetry bootstrap
|
||||
2. Add the Logfire `loguru` handler to the existing `loguru` configuration
|
||||
3. At operation boundaries, bind stable contextual fields with `loguru`
|
||||
4. Let logs emitted inside the span inherit the active trace context
|
||||
|
||||
### Context to bind
|
||||
|
||||
Bind only the fields that help correlate work across the system:
|
||||
|
||||
- `service_name`
|
||||
- `entrypoint`
|
||||
- `project_name`
|
||||
- `workspace_id`
|
||||
- `route_mode`
|
||||
- `tool_name`
|
||||
- `command_name`
|
||||
|
||||
This binding should happen at the root of an operation, not deep in leaf functions.
|
||||
|
||||
### Important nuance
|
||||
|
||||
We should not try to encode the entire trace model into logger extras. The logger context should be a human-meaningful slice of the active operation. Trace linkage comes from the active Logfire/OpenTelemetry context; logger extras are there to improve searchability and readability.
|
||||
|
||||
## Span Model
|
||||
|
||||
### Root spans
|
||||
|
||||
Each user-visible or system-visible operation should get one root span.
|
||||
|
||||
Examples:
|
||||
|
||||
- `cli.command.status`
|
||||
- `cli.command.project_sync`
|
||||
- `api.request.search`
|
||||
- `mcp.tool.write_note`
|
||||
- `mcp.tool.read_note`
|
||||
- `mcp.tool.search_notes`
|
||||
- `sync.project.run`
|
||||
- `db.semantic_backfill`
|
||||
|
||||
### Child spans
|
||||
|
||||
Child spans should represent real phases whose duration we care about.
|
||||
|
||||
Examples:
|
||||
|
||||
- `routing.client_session`
|
||||
- `routing.resolve_project`
|
||||
- `routing.resolve_workspace`
|
||||
- `api.search.execute`
|
||||
- `sync.project.scan`
|
||||
- `sync.project.detect_moves`
|
||||
- `sync.project.apply_changes`
|
||||
- `sync.project.resolve_relations`
|
||||
- `sync.project.sync_embeddings`
|
||||
- `sync.file.markdown`
|
||||
- `sync.file.regular`
|
||||
- `search.execute`
|
||||
- `search.relaxed_fts_retry`
|
||||
- `db.init`
|
||||
- `db.migrate`
|
||||
|
||||
### Span naming rules
|
||||
|
||||
- Use dot-separated names
|
||||
- Start with subsystem
|
||||
- Keep the verb at the end
|
||||
- Keep names stable across runs
|
||||
- Never include request-specific text in the span name
|
||||
|
||||
## Attribute Taxonomy
|
||||
|
||||
### Required attributes on root spans
|
||||
|
||||
Every root span should have a small common set:
|
||||
|
||||
- `service_name`
|
||||
- `entrypoint`
|
||||
- `project_name` when applicable
|
||||
- `workspace_id` when applicable
|
||||
- `route_mode` with values like `local_asgi`, `cloud_proxy`, `factory`
|
||||
|
||||
### Operation-specific attributes
|
||||
|
||||
Examples:
|
||||
|
||||
For search:
|
||||
|
||||
- `search_type`
|
||||
- `retrieval_mode`
|
||||
- `page`
|
||||
- `page_size`
|
||||
- `result_count`
|
||||
- `fallback_used`
|
||||
|
||||
For sync:
|
||||
|
||||
- `scan_type`
|
||||
- `force_full`
|
||||
- `new_count`
|
||||
- `modified_count`
|
||||
- `deleted_count`
|
||||
- `move_count`
|
||||
- `skipped_count`
|
||||
- `embeddings_enabled`
|
||||
|
||||
For note operations:
|
||||
|
||||
- `tool_name`
|
||||
- `note_type`
|
||||
- `directory`
|
||||
- `overwrite`
|
||||
- `output_format`
|
||||
|
||||
### Attributes to avoid by default
|
||||
|
||||
- full `query.text`
|
||||
- full note titles if they create privacy or cardinality issues
|
||||
- file content
|
||||
- raw frontmatter
|
||||
- raw HTTP bodies
|
||||
|
||||
If we need richer payloads for a local debugging session, that should be an explicit temporary mode, not the default telemetry shape.
|
||||
|
||||
## Instrumentation Plan By Layer
|
||||
|
||||
### 1. Entrypoints
|
||||
|
||||
Instrument these first:
|
||||
|
||||
- `cli.app` callback and major commands
|
||||
- API lifespan and selected routers
|
||||
- MCP server lifespan
|
||||
- MCP tool entrypoints
|
||||
|
||||
Why:
|
||||
|
||||
- this establishes clean root spans
|
||||
- it gives us trace boundaries that match how users think about the product
|
||||
|
||||
### 2. Routing and context resolution
|
||||
|
||||
Instrument:
|
||||
|
||||
- client routing decisions
|
||||
- workspace resolution
|
||||
- project resolution
|
||||
- default-project fallback
|
||||
|
||||
Why:
|
||||
|
||||
- Basic Memory has local/cloud/per-project routing logic
|
||||
- when something is slow or surprising, we need to know which path was taken
|
||||
|
||||
### 3. Sync and indexing
|
||||
|
||||
This is the highest-value area to instrument deeply.
|
||||
|
||||
Instrument:
|
||||
|
||||
- sync root
|
||||
- scan strategy decision
|
||||
- filesystem scan
|
||||
- move detection
|
||||
- delete handling
|
||||
- markdown sync phase
|
||||
- relation resolution
|
||||
- vector embedding sync
|
||||
- scan watermark update
|
||||
|
||||
Why:
|
||||
|
||||
- this is where performance work will happen
|
||||
- cloud and local both benefit from this visibility
|
||||
|
||||
### 4. Search
|
||||
|
||||
Instrument:
|
||||
|
||||
- search execution
|
||||
- retrieval mode
|
||||
- relaxed FTS fallback
|
||||
- result shaping
|
||||
|
||||
Why:
|
||||
|
||||
- search is user-facing and latency-sensitive
|
||||
- hybrid/vector/FTS paths need to be distinguishable
|
||||
|
||||
### 5. Database and initialization
|
||||
|
||||
Instrument selectively:
|
||||
|
||||
- DB init
|
||||
- migrations
|
||||
- semantic backfill
|
||||
- connection mode selection
|
||||
|
||||
Avoid full automatic SQL span firehose by default.
|
||||
|
||||
## Recommended Rollout Phases
|
||||
|
||||
## Task List
|
||||
|
||||
- [x] Phase 1: Bootstrap and config gating
|
||||
- [x] Phase 2: Root spans for entrypoints and primary operations
|
||||
- [x] Phase 3: Child spans for sync, search, and routing
|
||||
- [x] Phase 4: Failure-focused detail and final verification
|
||||
- [x] Phase 5: Loguru context binding and scoped context inheritance
|
||||
|
||||
## Recommended Rollout Phases
|
||||
|
||||
### Phase 1: Bootstrap and config gating
|
||||
|
||||
Add:
|
||||
|
||||
- telemetry bootstrap module
|
||||
- config/env gating
|
||||
- `loguru` + Logfire handler integration
|
||||
|
||||
This gives immediate value with low noise.
|
||||
|
||||
### Phase 2: Root spans for entrypoints and primary operations
|
||||
|
||||
Add:
|
||||
|
||||
- root spans for CLI, API, MCP, and main MCP tools
|
||||
- stable root attributes for project, workspace, route mode, and operation type
|
||||
|
||||
This gives us clean top-level traces that match how users think about the product.
|
||||
|
||||
### Phase 3: Child spans for sync, search, and routing
|
||||
|
||||
Add child spans to:
|
||||
|
||||
- sync
|
||||
- search
|
||||
- routing
|
||||
|
||||
This is the main performance-investigation layer.
|
||||
|
||||
### Phase 4: Failure-focused detail
|
||||
|
||||
Add selective deeper spans/log enrichment for:
|
||||
|
||||
- sync failures
|
||||
- relation resolution failures
|
||||
- slow file operations
|
||||
- cloud routing/auth failures
|
||||
|
||||
This keeps normal traces clean while improving debuggability.
|
||||
|
||||
### Phase 5: Loguru context binding and scoped context inheritance
|
||||
|
||||
Add:
|
||||
|
||||
- context-local telemetry state in `basic_memory.telemetry`
|
||||
- a shared `scope(...)` helper that opens a span and binds stable logger context together
|
||||
- context inheritance for routing, sync, and search so downstream `loguru` logs carry the active operation fields
|
||||
|
||||
This makes the trace view and the log stream tell the same story without forcing logger rewrites across the codebase.
|
||||
|
||||
## Local Dev Playbook
|
||||
|
||||
The fastest way to sanity-check the current trace shape is:
|
||||
|
||||
```bash
|
||||
LOGFIRE_TOKEN=lf_... just telemetry-smoke
|
||||
```
|
||||
|
||||
What this does:
|
||||
|
||||
- creates an isolated temp home, config dir, and project path
|
||||
- enables Logfire for the run
|
||||
- automatically exports to Logfire when `LOGFIRE_TOKEN` is present
|
||||
- defaults `BASIC_MEMORY_SEMANTIC_SEARCH_ENABLED=false` so the smoke run stays fast and trace-friendly
|
||||
- disables promo telemetry so the trace is about Basic Memory work, not analytics noise
|
||||
- runs a small CLI workflow:
|
||||
- `project add`
|
||||
- `tool write-note`
|
||||
- `tool read-note`
|
||||
- `tool edit-note`
|
||||
- `tool build-context`
|
||||
- `tool search-notes`
|
||||
- `doctor`
|
||||
|
||||
If you want to exercise the instrumentation without exporting anything upstream:
|
||||
|
||||
```bash
|
||||
BASIC_MEMORY_LOGFIRE_SEND_TO_LOGFIRE=false just telemetry-smoke
|
||||
```
|
||||
|
||||
If you want the smoke run to include vector or hybrid retrieval spans too:
|
||||
|
||||
```bash
|
||||
LOGFIRE_TOKEN=lf_... BASIC_MEMORY_SEMANTIC_SEARCH_ENABLED=true just telemetry-smoke
|
||||
```
|
||||
|
||||
The recipe sets `BASIC_MEMORY_LOGFIRE_ENVIRONMENT=telemetry-smoke` by default so these traces are easy to isolate in Logfire. Override it if you want the smoke traces grouped under a different environment name.
|
||||
|
||||
### What to look for
|
||||
|
||||
You should see a small set of comparable root spans rather than a framework-generated span forest:
|
||||
|
||||
- `cli.command.project`
|
||||
- `cli.command.tool`
|
||||
- `mcp.tool.write_note`
|
||||
- `mcp.tool.read_note`
|
||||
- `mcp.tool.edit_note`
|
||||
- `mcp.tool.build_context`
|
||||
- `mcp.tool.search_notes`
|
||||
- `sync.project.run`
|
||||
|
||||
You should also see correlated logs under those traces with stable fields like:
|
||||
|
||||
- `project_name`
|
||||
- `route_mode`
|
||||
- `tool_name`
|
||||
- `entrypoint`
|
||||
|
||||
### Expected nuance
|
||||
|
||||
`doctor` creates its own temporary project on purpose. That means the sync trace will usually show a different project name than the `telemetry-smoke` write/search traces. That is fine for smoke testing because the goal is to confirm:
|
||||
|
||||
- root span names are meaningful
|
||||
- scoped logs stay attached to the active trace
|
||||
- routing, tool, search, and sync phases are easy to distinguish
|
||||
|
||||
## Validation Checklist
|
||||
|
||||
We should consider the integration successful when the following are true:
|
||||
|
||||
1. With telemetry disabled, Basic Memory behaves exactly as it does today.
|
||||
2. With telemetry enabled, one user action produces one obvious root span.
|
||||
3. Logs emitted during that action are visible inside the same trace.
|
||||
4. A search in Logfire for `mcp.tool.write_note` or `sync.project.run` returns comparable spans across runs.
|
||||
5. Trace views show phase timing clearly without drowning in framework noise.
|
||||
6. Sensitive payloads are not captured by default.
|
||||
|
||||
## Immediate Implementation Direction
|
||||
|
||||
When we start coding, the first pass should be:
|
||||
|
||||
1. Add `basic_memory.telemetry`
|
||||
2. Add config/env switches for `enabled`, `send_to_logfire`, and service name
|
||||
3. Wire telemetry bootstrap into CLI, API, and MCP entrypoints
|
||||
4. Configure `loguru` to emit to both existing sinks and the Logfire handler when enabled
|
||||
5. Add manual root spans around:
|
||||
- CLI commands
|
||||
- API request handlers we care about
|
||||
- MCP tool entrypoints
|
||||
- sync root
|
||||
- search root
|
||||
6. Add child spans to the sync and routing phases only after the root span model feels clean
|
||||
|
||||
That gives us a strong foundation without repeating the earlier "turn on instrumentation everywhere" approach.
|
||||
@@ -1,183 +0,0 @@
|
||||
# Manual Pages
|
||||
|
||||
Basic Memory's manual is written in the style of Unix man pages — and
|
||||
implemented as Basic Memory notes ([#952](https://github.com/basicmachines-co/basic-memory/issues/952)).
|
||||
Every page is a markdown note conforming to the `Manpage` schema, `SEE ALSO`
|
||||
entries are real knowledge-graph relations, and every example on every page
|
||||
was executed against a live project before the page shipped. The manual
|
||||
documents the tools; the tools verify the manual.
|
||||
|
||||
## Where it lives
|
||||
|
||||
The canonical manual is the **`manual` project in the Basic Memory team
|
||||
workspace** (cloud, shared). Anyone can build their own: the schema ships as
|
||||
an opt-in seed at `plugins/claude-code/schemas/manpage.md` — copy it into any
|
||||
project's folder and start writing pages against it.
|
||||
|
||||
Layout:
|
||||
|
||||
```
|
||||
manual/
|
||||
├── schemas/Manpage.md # the manpage schema (type: schema)
|
||||
├── man1/ # CLI commands bm(1), bm-status(1), ...
|
||||
├── man3/ # MCP tools write-note(3), search-notes(3), ...
|
||||
├── man5/ # file formats bm-note(5), bm-observation(5), ...
|
||||
├── man7/ # concepts basic-memory(7), semantic-memory(7), ...
|
||||
├── playground/ # scratch notes for destructive examples
|
||||
└── diagrams/ # canvas visualizations of the manual graph
|
||||
```
|
||||
|
||||
### Why "man1", "man3", "man5"?
|
||||
|
||||
The folder names are Unix's, unchanged since 1971. The manual is divided
|
||||
into numbered **sections**, pages physically live in directories named
|
||||
after them (`/usr/share/man/man1`, `man5`, ...), and the number tells you
|
||||
what *kind* of thing is documented — not importance, not reading order:
|
||||
|
||||
- **1** — user commands (`ls`, `grep`)
|
||||
- **2** — system calls
|
||||
- **3** — library functions / APIs (`printf(3)`)
|
||||
- **4** — devices
|
||||
- **5** — file formats and config files (`crontab(5)`, `passwd(5)`)
|
||||
- **6** — games (really)
|
||||
- **7** — miscellanea: concepts, conventions, overviews (`regex(7)`, `signal(7)`)
|
||||
- **8** — system administration
|
||||
|
||||
That's also why man page names carry the parenthesized number —
|
||||
`crontab(1)` is the command, `crontab(5)` is the file format, same name in
|
||||
two sections. `man 5 crontab` picks the section explicitly.
|
||||
|
||||
This manual copies that layout with the sections that have a Basic Memory
|
||||
analog:
|
||||
|
||||
- **man1/** — `bm` CLI commands → `bm-status(1)`
|
||||
- **man3/** — MCP tools, our equivalent of the "library API" section → `write-note(3)`
|
||||
- **man5/** — file formats: note syntax, observations, relations, schemas → `bm-note(5)`
|
||||
- **man7/** — concepts → `basic-memory(7)`, `semantic-memory(7)`
|
||||
- **8** is reserved for admin/cloud operations but has no pages yet; 2, 4,
|
||||
and 6 have no analog (no system calls, no devices, and no games — yet)
|
||||
|
||||
When a page says `see_also [[bm-note(5)]]`, the `(5)` reads "the
|
||||
file-format page," exactly the way a Unix manual cross-references — except
|
||||
here it's a traversable relation in the graph instead of a typographic
|
||||
convention. The manual explains its own conventions in `man-pages(7)` —
|
||||
fittingly, the same page name Linux uses for this, and that almost nobody
|
||||
ever reads.
|
||||
|
||||
## Page anatomy
|
||||
|
||||
Pages use the classic headers where applicable: `NAME`, `SYNOPSIS`,
|
||||
`DESCRIPTION`, `PARAMETERS`, `MCP USAGE`, `CLI EQUIVALENT`, `EXAMPLES`,
|
||||
`GOTCHAS`, `SEE ALSO`. Frontmatter (validated by the schema):
|
||||
|
||||
```yaml
|
||||
type: manpage
|
||||
section: 3 # 1 | 3 | 5 | 7 | 8
|
||||
name: write-note # page name without section suffix
|
||||
summary: create or overwrite a markdown note in the knowledge base
|
||||
generated: hand # hand | registry | typer (regeneration ownership)
|
||||
tool: write_note # section-3 pages: the MCP tool documented
|
||||
command: basic-memory status # section-1 pages: the CLI command documented
|
||||
verified: 0.21.6 mcp+cli # version + path(s) that proved the page
|
||||
```
|
||||
|
||||
Field knowledge accumulates as observations — `[gotcha]`, `[bug]` (with issue
|
||||
links), `[pattern]` — and `SEE ALSO` entries are `see_also` relations, so the
|
||||
manual is a navigable graph, not a folder of files.
|
||||
|
||||
## How to use it
|
||||
|
||||
Man-style reads (any MCP client or the CLI):
|
||||
|
||||
```bash
|
||||
# read a page
|
||||
bm tool read-note "man3/write-note-3" --project manual
|
||||
|
||||
# apropos — find pages by section, tool, or text
|
||||
bm tool search-notes --project manual # then filter, or via MCP:
|
||||
# search_notes(project="manual", metadata_filters={"type": "manpage", "section": 3})
|
||||
# search_notes(project="manual", metadata_filters={"type": "manpage", "tool": "write_note"})
|
||||
|
||||
# traverse SEE ALSO from any page
|
||||
# build_context(url="man3/write-note-3", project="manual")
|
||||
```
|
||||
|
||||
A future `bm man <topic>` command is thin sugar over exactly these calls.
|
||||
|
||||
And for the real thing — `man bm` in an actual terminal:
|
||||
|
||||
```bash
|
||||
bm man install # copies bundled groff pages to ~/.local/share/man
|
||||
man bm # the overview page, rendered by man(1)
|
||||
man basic-memory # same page via its alias
|
||||
```
|
||||
|
||||
`bm man install` warns with a one-line `MANPATH` fix if the install root
|
||||
isn't searched by your `man`. Agents with shell access can use `man bm` as
|
||||
an offline quick reference; the full per-tool detail stays in the manual
|
||||
project's section-3 pages.
|
||||
|
||||
## The verification discipline
|
||||
|
||||
Two rules make the manual trustworthy:
|
||||
|
||||
1. **Examples must have run.** An `EXAMPLES` (or `MCP USAGE` / `CLI
|
||||
EQUIVALENT`) block contains only commands that actually executed against
|
||||
the manual project. Destructive operations (`delete_note`, `move_note`,
|
||||
destructive `edit_note`) run only against `playground/` notes — never
|
||||
against pages. The `verified:` field records the version and which path
|
||||
proved the page: `mcp` (live service), `cli` (dev checkout), or both.
|
||||
|
||||
2. **The schema is the linter.** Validate the whole manual any time:
|
||||
|
||||
```bash
|
||||
bm tool schema-validate manpage --project manual
|
||||
# → {"total_notes": 38, "valid_count": 38, "warning_count": 0, ...}
|
||||
```
|
||||
|
||||
`bm orphans --project manual` confirms every page is connected to the
|
||||
graph, and `schema_diff`/`schema_infer` report drift between the schema
|
||||
and how pages are actually written.
|
||||
|
||||
Because verification exercises real tool calls against the live service,
|
||||
building the manual doubles as an end-to-end smoke test. The initial build
|
||||
found six bugs in one pass (#954–#959) — including the verification rule
|
||||
catching a test that asserted a bug as expected output (#958).
|
||||
|
||||
## Adding or updating a page
|
||||
|
||||
1. Run the commands you intend to document; keep the actual output.
|
||||
2. Write the page with `write_note`, passing frontmatter through the
|
||||
`metadata` parameter (nested YAML in content frontmatter is unreliable on
|
||||
some clients):
|
||||
|
||||
```
|
||||
write_note(title="my-tool(3)", directory="man3", project="manual",
|
||||
note_type="manpage",
|
||||
metadata={"section": 3, "name": "my-tool",
|
||||
"summary": "...", "generated": "hand",
|
||||
"tool": "my_tool", "verified": "<version> mcp"})
|
||||
```
|
||||
|
||||
3. Link related pages in `SEE ALSO` with `see_also [[other-page(3)]]`.
|
||||
Forward references to pages that don't exist yet are fine — they resolve
|
||||
automatically when the target is written.
|
||||
4. Validate: `bm tool schema-validate manpage --project manual`.
|
||||
|
||||
For mechanical updates to generated sections, prefer `edit_note` with
|
||||
`replace_section` / `insert_after_section` so curated content (EXAMPLES,
|
||||
GOTCHAS, SEE ALSO, observations) survives — that ownership split is what the
|
||||
`generated:` field declares.
|
||||
|
||||
## Roadmap
|
||||
|
||||
- **Registry generator** — section-3 SYNOPSIS/PARAMETERS generated from the
|
||||
MCP tool registry (docstrings + pydantic schemas), section-1 from Typer
|
||||
help; the hand-written corpus is the template spec. Regenerate-and-diff in
|
||||
CI becomes the drift gate.
|
||||
- **`bm man <topic>`** — CLI sugar over `read_note` + metadata search.
|
||||
(`bm man install` + a hand-written `bm.1` already ship — the first slice
|
||||
of [#610](https://github.com/basicmachines-co/basic-memory/issues/610);
|
||||
the generator will produce per-command pages from the same extraction.)
|
||||
- **Docs site** — the notes remain canonical for sections 5 and 7, code is
|
||||
canonical for 1 and 3; both render to the hosted docs site.
|
||||
@@ -1,138 +0,0 @@
|
||||
# MCP UI Bakeoff - Instructions & Test Plan
|
||||
|
||||
Last updated: 2026-02-02
|
||||
|
||||
## Scope
|
||||
|
||||
Compare three presentation paths for Basic Memory MCP tools:
|
||||
|
||||
1. **Tool‑UI (React)** via MCP App resources.
|
||||
2. **MCP‑UI Python SDK** embedded UI resources (legacy host path).
|
||||
3. **ASCII/ANSI** output for TUI clients.
|
||||
|
||||
This doc is the running instruction set and test plan. Update as implementation progresses.
|
||||
|
||||
---
|
||||
|
||||
## Prerequisites
|
||||
|
||||
- Repo: `basic-memory` (worktree: `basic-memory-mcp-ui-poc`)
|
||||
- Node for tool‑ui build (already used for POC)
|
||||
- Python 3.12+ with `uv`
|
||||
|
||||
Optional (for MCP‑UI Python SDK path):
|
||||
|
||||
- Local repo: `/Users/phernandez/dev/mcp-ui`
|
||||
- Install the server SDK into the Basic Memory venv:
|
||||
- `uv pip install -e /Users/phernandez/dev/mcp-ui/sdks/python/server`
|
||||
|
||||
---
|
||||
|
||||
## Build / Refresh Steps
|
||||
|
||||
### Tool‑UI React bundle
|
||||
|
||||
```bash
|
||||
cd ui/tool-ui-react
|
||||
npm install
|
||||
npm run build
|
||||
```
|
||||
|
||||
This regenerates:
|
||||
|
||||
- `src/basic_memory/mcp/ui/html/search-results-tool-ui.html`
|
||||
- `src/basic_memory/mcp/ui/html/note-preview-tool-ui.html`
|
||||
|
||||
---
|
||||
|
||||
## How to Run the MCP Server
|
||||
|
||||
```bash
|
||||
basic-memory mcp --transport stdio
|
||||
```
|
||||
|
||||
Optional to pick UI variant for MCP App resources:
|
||||
|
||||
```bash
|
||||
export BASIC_MEMORY_MCP_UI_VARIANT=tool-ui # or vanilla | mcp-ui
|
||||
```
|
||||
|
||||
---
|
||||
|
||||
## Test Cases
|
||||
|
||||
### 1) MCP App Resource UI (tool‑ui / vanilla / mcp‑ui)
|
||||
|
||||
Tools:
|
||||
- `search_notes`
|
||||
- `read_note`
|
||||
|
||||
Expect:
|
||||
- Tool meta points to `ui://basic-memory/search-results` and `ui://basic-memory/note-preview`
|
||||
- Resource content differs by `BASIC_MEMORY_MCP_UI_VARIANT`
|
||||
- Variant‑specific URIs also available:
|
||||
- `ui://basic-memory/search-results/vanilla`
|
||||
- `ui://basic-memory/search-results/tool-ui`
|
||||
- `ui://basic-memory/search-results/mcp-ui`
|
||||
- `ui://basic-memory/note-preview/vanilla`
|
||||
- `ui://basic-memory/note-preview/tool-ui`
|
||||
- `ui://basic-memory/note-preview/mcp-ui`
|
||||
|
||||
Manual check:
|
||||
- Trigger tool in MCP‑App‑capable host and confirm UI renders.
|
||||
|
||||
---
|
||||
|
||||
### 2) Text / JSON Output Modes
|
||||
|
||||
Tools:
|
||||
- `search_notes(output_format="text" | "json")`
|
||||
- `read_note(output_format="text" | "json")`
|
||||
- `write_note(output_format="text" | "json")`
|
||||
- `edit_note(output_format="text" | "json")`
|
||||
- `recent_activity(output_format="text" | "json")`
|
||||
- `list_memory_projects(output_format="text" | "json")`
|
||||
- `create_memory_project(output_format="text" | "json")`
|
||||
- `delete_note(output_format="text" | "json")`
|
||||
- `move_note(output_format="text" | "json")`
|
||||
- `build_context(output_format="json" | "text")`
|
||||
|
||||
Expect:
|
||||
- `text` mode preserves existing human-readable responses.
|
||||
- `json` mode returns structured dict/list payloads for machine-readable clients.
|
||||
|
||||
Automated:
|
||||
- `uv run pytest test-int/mcp/test_output_format_json_integration.py`
|
||||
|
||||
---
|
||||
|
||||
### 3) MCP‑UI Python SDK (embedded UI resource)
|
||||
|
||||
Tools (embedded resource responses):
|
||||
- `search_notes_ui` (MCP‑UI SDK)
|
||||
- `read_note_ui` (MCP‑UI SDK)
|
||||
|
||||
Expected output:
|
||||
- Tool response content contains an EmbeddedResource (`type: "resource"`)
|
||||
- `mimeType` is `text/html`
|
||||
- `_meta` includes:
|
||||
- `mcpui.dev/ui-preferred-frame-size`
|
||||
- `mcpui.dev/ui-initial-render-data`
|
||||
|
||||
Manual check:
|
||||
- Render tool responses using `UIResourceRenderer` (legacy host flow).
|
||||
|
||||
Automated (if SDK installed):
|
||||
- `uv run pytest test-int/mcp/test_ui_sdk_integration.py`
|
||||
|
||||
---
|
||||
|
||||
## Bakeoff Notes Template
|
||||
|
||||
Fill in after running:
|
||||
|
||||
- Tool‑UI (React): __
|
||||
- MCP‑UI SDK (embedded): __
|
||||
- Text/JSON modes: __
|
||||
|
||||
Decision + rationale: __
|
||||
@@ -1,260 +0,0 @@
|
||||
# Metadata Search Reference
|
||||
|
||||
Basic Memory automatically indexes custom frontmatter fields so you can query them with structured filters. Any YAML key in a note's frontmatter beyond the standard set (`title`, `type`, `tags`, `permalink`, `schema`) is stored as `entity_metadata` and becomes searchable.
|
||||
|
||||
## Querying with `search_notes`
|
||||
|
||||
`search_notes` is the single search tool for all queries — text, metadata filters, or both. The `query` parameter is optional, so you can use metadata filters alone without passing an empty string.
|
||||
|
||||
## Filter Syntax
|
||||
|
||||
Filters are a JSON dictionary where each key targets a frontmatter field and the value specifies the match condition. Multiple keys combine with **AND** logic — every filter must match.
|
||||
|
||||
### Equality
|
||||
|
||||
Match a single value exactly.
|
||||
|
||||
```json
|
||||
{"status": "active"}
|
||||
```
|
||||
|
||||
Finds notes whose frontmatter contains `status: active`.
|
||||
|
||||
### Array Contains (all)
|
||||
|
||||
Pass a list to require **all** listed values to be present in the field.
|
||||
|
||||
```json
|
||||
{"tags": ["security", "oauth"]}
|
||||
```
|
||||
|
||||
Finds notes tagged with both `security` and `oauth`.
|
||||
|
||||
### `$in` (any of)
|
||||
|
||||
Match if the field equals **any** value in the list.
|
||||
|
||||
```json
|
||||
{"priority": {"$in": ["high", "critical"]}}
|
||||
```
|
||||
|
||||
### `$gt`, `$gte`, `$lt`, `$lte`
|
||||
|
||||
Numeric and text comparisons. Numeric values use numeric comparison; strings use lexicographic comparison.
|
||||
|
||||
```json
|
||||
{"confidence": {"$gt": 0.7}}
|
||||
{"score": {"$lte": 100}}
|
||||
```
|
||||
|
||||
### `$between`
|
||||
|
||||
Range filter (inclusive). Takes a `[min, max]` pair.
|
||||
|
||||
```json
|
||||
{"score": {"$between": [0.3, 0.8]}}
|
||||
```
|
||||
|
||||
### Nested Access (dot notation)
|
||||
|
||||
Access nested frontmatter values using dots.
|
||||
|
||||
```json
|
||||
{"schema.version": "2"}
|
||||
```
|
||||
|
||||
This queries the `version` key inside a `schema` object in frontmatter.
|
||||
|
||||
### Summary Table
|
||||
|
||||
| Operator | Syntax | Example |
|
||||
|----------|--------|---------|
|
||||
| Equality | `{"field": "value"}` | `{"status": "active"}` |
|
||||
| Array contains (all) | `{"field": ["a", "b"]}` | `{"tags": ["security", "oauth"]}` |
|
||||
| `$in` (any of) | `{"field": {"$in": [...]}}` | `{"priority": {"$in": ["high", "critical"]}}` |
|
||||
| `$gt` / `$gte` | `{"field": {"$gt": N}}` | `{"confidence": {"$gt": 0.7}}` |
|
||||
| `$lt` / `$lte` | `{"field": {"$lt": N}}` | `{"score": {"$lt": 0.5}}` |
|
||||
| `$between` | `{"field": {"$between": [min, max]}}` | `{"score": {"$between": [0.3, 0.8]}}` |
|
||||
| Nested access | `{"a.b": "value"}` | `{"schema.version": "2"}` |
|
||||
|
||||
**Key rules:**
|
||||
- Filter keys must match `[A-Za-z0-9_-]+` (dots separate nesting levels).
|
||||
- Each operator dict must contain exactly one operator.
|
||||
- `$in` and array-contains require non-empty lists.
|
||||
- `$between` requires exactly two values `[min, max]`.
|
||||
|
||||
## MCP Tool — `search_notes`
|
||||
|
||||
`search_notes` is the single search tool for text queries, metadata filters, or both. The `query` parameter is optional.
|
||||
|
||||
**Relevant parameters:**
|
||||
|
||||
| Parameter | Type | Description |
|
||||
|-----------|------|-------------|
|
||||
| `query` | string (optional) | Text search query. Omit for filter-only searches. |
|
||||
| `metadata_filters` | dict | Structured filter dict (see syntax above) |
|
||||
| `tags` | list[str] | Convenience shorthand — merged into `metadata_filters["tags"]` |
|
||||
| `status` | string | Convenience shorthand — merged into `metadata_filters["status"]` |
|
||||
|
||||
**Merging rules:** `tags` and `status` are convenience shortcuts. They are merged into `metadata_filters` using `setdefault` — if the same key already exists in `metadata_filters`, the explicit filter wins.
|
||||
|
||||
**Examples:**
|
||||
|
||||
```python
|
||||
# Text search filtered by metadata
|
||||
await search_notes("authentication", metadata_filters={"status": "draft"})
|
||||
|
||||
# Filter-only search (no query needed)
|
||||
await search_notes(metadata_filters={"type": "spec"})
|
||||
|
||||
# Combine text, tags shortcut, and metadata
|
||||
await search_notes(
|
||||
"oauth flow",
|
||||
tags=["security"],
|
||||
metadata_filters={"confidence": {"$gt": 0.7}},
|
||||
)
|
||||
|
||||
# Convenience shortcuts
|
||||
await search_notes("planning", status="active")
|
||||
await search_notes(tags=["tier1", "alpha"])
|
||||
```
|
||||
|
||||
## Tag Search Shortcuts
|
||||
|
||||
The `tag:` prefix in a search query is a shorthand for tag-based metadata filtering. When `search_notes` receives a query starting with `tag:`, it converts the query into a `tags` filter and clears the text query.
|
||||
|
||||
```python
|
||||
# These are equivalent:
|
||||
await search_notes("tag:tier1")
|
||||
await search_notes("", tags=["tier1"])
|
||||
|
||||
# Multiple tags (comma or space separated) — all must be present:
|
||||
await search_notes("tag:tier1,alpha")
|
||||
await search_notes("tag:tier1 alpha")
|
||||
```
|
||||
|
||||
## CLI Access
|
||||
|
||||
The `bm tool search-notes` command exposes metadata filtering via `--meta` and `--filter` flags.
|
||||
|
||||
### `--meta` — simple key=value filters
|
||||
|
||||
Repeatable flag for equality filters on frontmatter fields.
|
||||
|
||||
```bash
|
||||
# Single filter
|
||||
bm tool search-notes "my query" --meta status=draft
|
||||
|
||||
# Multiple filters (AND logic)
|
||||
bm tool search-notes "" --meta status=active --meta priority=high
|
||||
```
|
||||
|
||||
### `--filter` — advanced JSON filters
|
||||
|
||||
Pass a full JSON filter dictionary for operator-based queries.
|
||||
|
||||
```bash
|
||||
# Range filter
|
||||
bm tool search-notes "" --filter '{"score": {"$between": [0.3, 0.8]}}'
|
||||
|
||||
# $in filter
|
||||
bm tool search-notes "" --filter '{"priority": {"$in": ["high", "critical"]}}'
|
||||
```
|
||||
|
||||
### `--tag` and `--status` — convenience shortcuts
|
||||
|
||||
```bash
|
||||
bm tool search-notes "query" --tag security --tag oauth
|
||||
bm tool search-notes "" --status draft
|
||||
```
|
||||
|
||||
### Combined example
|
||||
|
||||
```bash
|
||||
bm tool search-notes "authentication" --tag security --meta status=draft --type spec
|
||||
```
|
||||
|
||||
## Practical Examples
|
||||
|
||||
### Example notes with custom frontmatter
|
||||
|
||||
**`specs/auth-design.md`:**
|
||||
|
||||
```markdown
|
||||
---
|
||||
title: Auth Design
|
||||
type: spec
|
||||
tags: [security, oauth]
|
||||
status: in-progress
|
||||
priority: high
|
||||
confidence: 0.85
|
||||
---
|
||||
|
||||
# Auth Design
|
||||
|
||||
## Observations
|
||||
- [decision] Use OAuth 2.1 with PKCE for all client types #security
|
||||
- [requirement] Token refresh must be transparent to the user
|
||||
|
||||
## Relations
|
||||
- implements [[Security Requirements]]
|
||||
```
|
||||
|
||||
**`specs/search-redesign.md`:**
|
||||
|
||||
```markdown
|
||||
---
|
||||
title: Search Redesign
|
||||
type: spec
|
||||
tags: [search, performance]
|
||||
status: draft
|
||||
priority: medium
|
||||
confidence: 0.6
|
||||
---
|
||||
|
||||
# Search Redesign
|
||||
|
||||
## Observations
|
||||
- [goal] Sub-100ms search response times #performance
|
||||
- [approach] Hybrid FTS + vector retrieval
|
||||
|
||||
## Relations
|
||||
- depends_on [[Database Schema]]
|
||||
```
|
||||
|
||||
### Queries that find them
|
||||
|
||||
```python
|
||||
# Find all in-progress specs
|
||||
await search_notes(metadata_filters={"status": "in-progress", "type": "spec"})
|
||||
# → Auth Design
|
||||
|
||||
# Find high-confidence specs
|
||||
await search_notes(metadata_filters={"confidence": {"$gt": 0.7}})
|
||||
# → Auth Design (confidence: 0.85)
|
||||
|
||||
# Find specs with priority high or medium
|
||||
await search_notes(metadata_filters={"priority": {"$in": ["high", "medium"]}})
|
||||
# → Auth Design, Search Redesign
|
||||
|
||||
# Find specs in a confidence range
|
||||
await search_notes(metadata_filters={"confidence": {"$between": [0.5, 0.9]}})
|
||||
# → Auth Design (0.85), Search Redesign (0.6)
|
||||
|
||||
# Find notes tagged with security
|
||||
await search_notes("tag:security")
|
||||
# → Auth Design
|
||||
|
||||
# Combined: text search + metadata filter
|
||||
await search_notes("OAuth", metadata_filters={"status": "in-progress"})
|
||||
# → Auth Design
|
||||
```
|
||||
|
||||
### CLI equivalents
|
||||
|
||||
```bash
|
||||
bm tool search-notes "" --meta status=in-progress --type spec
|
||||
bm tool search-notes "" --filter '{"confidence": {"$gt": 0.7}}'
|
||||
bm tool search-notes "OAuth" --meta status=in-progress
|
||||
bm tool search-notes --tag security
|
||||
```
|
||||
@@ -1,344 +0,0 @@
|
||||
# Post-v0.18.0 Test Plan and Acceptance Criteria
|
||||
|
||||
## Goal
|
||||
|
||||
Define a complete validation plan for all major features merged after `v0.18.0`, combining:
|
||||
|
||||
- Coverage-gap-driven automated tests
|
||||
- Real MCP server integration tests (no mocks for target flows)
|
||||
- Manual MCP verification via LLM-driven tool calls
|
||||
|
||||
This plan is based on commits in `v0.18.0..HEAD` and the latest `just check` coverage output.
|
||||
|
||||
## Scope Window
|
||||
|
||||
- Start tag: `v0.18.0` (2026-01-28)
|
||||
- End: current `main`
|
||||
- Change volume: 12 feature commits + 14 bug-fix commits (+ release chores/hotfixes)
|
||||
|
||||
## Execution Strategy
|
||||
|
||||
1. Stabilize all feature-level acceptance criteria in automated tests first.
|
||||
2. Add black-box MCP integration tests for semantic search + schema (real server startup).
|
||||
3. Run manual MCP tool-call verification to confirm real UX and routing behavior.
|
||||
4. Re-run full gate: `just check` + targeted integration packs.
|
||||
|
||||
## Global Quality Gates
|
||||
|
||||
- Feature criteria below must all pass.
|
||||
- No regressions in existing suites.
|
||||
- Coverage improves in targeted low-coverage feature modules.
|
||||
- SQLite and Postgres parity for search/semantic features.
|
||||
|
||||
## Priority Coverage Gaps (from latest run)
|
||||
|
||||
These are the most important post-`v0.18.0` feature modules currently under-covered:
|
||||
|
||||
- `src/basic_memory/mcp/tools/schema.py` (27%)
|
||||
- `src/basic_memory/mcp/clients/schema.py` (36%)
|
||||
- `src/basic_memory/mcp/tools/ui_sdk.py` (43%)
|
||||
- `src/basic_memory/mcp/tools/search.py` (73%)
|
||||
- `src/basic_memory/repository/postgres_search_repository.py` (63%)
|
||||
- `src/basic_memory/mcp/async_client.py` (82%)
|
||||
- `src/basic_memory/api/v2/routers/schema_router.py` (80%)
|
||||
|
||||
## Feature Acceptance Criteria and Test Plan
|
||||
|
||||
### 1) Schema System (`c97733d`) — DONE
|
||||
|
||||
### Acceptance criteria
|
||||
|
||||
- `schema_validate`, `schema_infer`, and `schema_diff` produce consistent outcomes across CLI/API/MCP for the same fixture set.
|
||||
- Strict validation fails deterministically on required-field/type violations.
|
||||
- Validation warnings are stable and machine-readable in non-strict mode.
|
||||
- Inference output is deterministic for unchanged input corpus.
|
||||
- Drift diff output is deterministic and identifies missing/extra/type-mismatch fields correctly.
|
||||
|
||||
### Existing coverage anchor points
|
||||
|
||||
- `tests/schema/*`
|
||||
- `tests/api/v2/test_schema_router.py`
|
||||
- `test-int/test_schema/*`
|
||||
|
||||
### Gaps to close — DONE
|
||||
|
||||
- ~~MCP schema tool branches (`src/basic_memory/mcp/tools/schema.py`)~~ — 18 tests in `tests/mcp/test_tool_schema.py`
|
||||
- ~~MCP schema client behavior (`src/basic_memory/mcp/clients/schema.py`)~~ — `tests/mcp/test_client_schema.py`
|
||||
- ~~Schema router error-path branches (`src/basic_memory/api/v2/routers/schema_router.py`)~~ — `tests/api/v2/test_schema_router.py`
|
||||
|
||||
### Planned additions — DONE
|
||||
|
||||
- ~~Add MCP tool tests for `schema_validate` strict + non-strict result shapes.~~ **DONE**
|
||||
- ~~Add MCP tool tests for `schema_infer` with explicit `entity_type` and inferred type fallback.~~ **DONE**
|
||||
- ~~Add MCP tool tests for `schema_diff` empty-diff and non-empty-diff paths.~~ **DONE**
|
||||
- ~~Add API tests for schema router invalid payload/edge error handling.~~ **DONE**
|
||||
- Add integration test that starts MCP server and calls schema tools end-to-end on fixture notes. — deferred to backlog item 4.
|
||||
|
||||
### 2) Semantic Search (`0777879`, `1428d18`, `344e651`) — DONE
|
||||
|
||||
### Acceptance criteria
|
||||
|
||||
- `search_type=text|vector|hybrid` returns expected ranked results on canonical semantic corpus.
|
||||
- Missing semantic dependencies fail fast with actionable install guidance.
|
||||
- Reindex and provider/model changes produce valid vectors without dimension mismatch.
|
||||
- SQLite and Postgres produce equivalent behavior for semantic modes on the same dataset.
|
||||
- Generated-column migration path is valid on SQLite environments in use.
|
||||
|
||||
### Existing coverage anchor points
|
||||
|
||||
- `tests/repository/test_sqlite_vector_search_repository.py`
|
||||
- `tests/repository/test_postgres_search_repository.py`
|
||||
- `tests/services/test_semantic_search.py`
|
||||
- `tests/mcp/test_tool_search.py`
|
||||
- `test-int/test_search_performance_benchmark.py`
|
||||
|
||||
### Gaps to close — DONE
|
||||
|
||||
- ~~Uncovered Postgres vector/hybrid branches~~ — 20 tests in `tests/repository/test_postgres_search_repository_unit.py` + 5 integration tests in `test-int/semantic/test_semantic_coverage.py`
|
||||
- ~~MCP search semantic/output branches~~ — expanded `tests/mcp/test_tool_search.py`
|
||||
|
||||
### Planned additions — DONE
|
||||
|
||||
- ~~Expand Postgres repository tests for vector query composition edge cases.~~ **DONE**
|
||||
- ~~Expand Postgres repository tests for hybrid fusion ranking and pagination branches.~~ **DONE**
|
||||
- ~~Expand Postgres repository tests for embedding/provider error handling branches.~~ **DONE**
|
||||
- ~~Expand MCP search tool tests for vector/hybrid output formatting branches.~~ **DONE**
|
||||
- ~~Expand MCP search tool tests for semantic-disabled and missing-dependency failures.~~ **DONE**
|
||||
- Add MCP integration tests that start server and execute semantic `search_notes` tool calls. — deferred to backlog item 4.
|
||||
|
||||
### Semantic search quality benchmarks (NEW)
|
||||
|
||||
Full benchmark suite in `test-int/semantic/` covering 5 backend×provider combinations:
|
||||
- `sqlite-fts`, `sqlite-fastembed`, `postgres-fts`, `postgres-fastembed`, `postgres-openai`
|
||||
- Quality metrics: hit@1, recall@5, MRR@10 with per-query timing
|
||||
- Realistic corpus with cross-topic vocabulary overlap (240 notes, 4 topics)
|
||||
- Rich CLI viewer: `just semantic-report`
|
||||
- JSON artifact output: `just test-semantic-report`
|
||||
|
||||
Key finding: **FastEmbed (384-d local ONNX) matches or exceeds OpenAI (1536-d) quality at 30x lower latency.** Recommending FastEmbed as default for both local and cloud deployments.
|
||||
|
||||
### 3) Per-Project Local/Cloud Routing + API Key Auth (`d84708c`, `ed94877`, `312662f`) — DONE
|
||||
|
||||
### Acceptance criteria
|
||||
|
||||
- Project mode (`local`/`cloud`) persists and displays correctly.
|
||||
- Routing selects ASGI for local projects and HTTP+Bearer for cloud projects.
|
||||
- Cloud project without key fails with explicit remediation (`cloud set-key`/`cloud create-key`).
|
||||
- Resolution precedence is correct (factory > force-local > per-project cloud > global fallback > local).
|
||||
- Watch/sync only run for local projects.
|
||||
|
||||
### Existing coverage anchor points
|
||||
|
||||
- `tests/mcp/test_async_client_modes.py`
|
||||
- `tests/cli/test_project_set_cloud_local.py`
|
||||
- `tests/mcp/test_project_context.py`
|
||||
- `tests/test_project_resolver.py`
|
||||
- `tests/sync/test_watch_service_reload.py`
|
||||
|
||||
### Gaps to close — DONE
|
||||
|
||||
- ~~Cloud routing branch gaps in `src/basic_memory/mcp/async_client.py`~~ — expanded `tests/mcp/test_async_client_modes.py`
|
||||
|
||||
### Planned additions — DONE
|
||||
|
||||
- ~~Add branch-focused tests for all unresolved routing branches in `get_client()`.~~ **DONE**
|
||||
- Add MCP integration scenario with mixed local/cloud project config — deferred to backlog item 4.
|
||||
|
||||
### 4) Project-Prefixed Permalinks + Memory URL Routing (`545804f`) — DONE
|
||||
|
||||
### Acceptance criteria
|
||||
|
||||
- Project-prefixed permalinks are generated consistently on create/update/import flows.
|
||||
- Memory URLs resolve to the correct project/entity even with duplicate note titles.
|
||||
- `read_note`, `search`, `build_context`, write/edit/move flows preserve project identity correctly.
|
||||
- Link resolution remains correct for context-aware wikilinks.
|
||||
|
||||
### Existing coverage anchor points
|
||||
|
||||
- `tests/utils/test_permalink_formatting.py`
|
||||
- `tests/mcp/test_tool_read_note.py`
|
||||
- `tests/mcp/test_tool_search.py`
|
||||
- `tests/services/test_context_service.py`
|
||||
- `test-int/mcp/test_read_note_integration.py`
|
||||
|
||||
### Gaps to close
|
||||
|
||||
- No major coverage alarm in report, but keep as regression-critical due broad impact surface.
|
||||
|
||||
### Planned additions — DONE
|
||||
|
||||
- ~~Add one integration test with colliding titles across two projects and assert URL routing invariants.~~ **DONE** — `test-int/mcp/test_permalink_collision_integration.py` (2 tests: collision across projects + memory:// URL routing with project prefix)
|
||||
|
||||
### 5) MCP UI Variants + TUI Output (`8bc03d1`) — DONE
|
||||
|
||||
### Acceptance criteria
|
||||
|
||||
- UI resource variant selection (`tool-ui`, `vanilla`, `mcp-ui`) follows env configuration.
|
||||
- `search_notes` and `read_note` expose expected resource metadata for UI hosts.
|
||||
- `ascii`/`ansi` outputs are deterministic and stable for terminal clients.
|
||||
|
||||
### Existing coverage anchor points
|
||||
|
||||
- `tests/mcp/test_tool_contracts.py`
|
||||
- `test-int/mcp/test_output_format_json_integration.py`
|
||||
- `test-int/mcp/test_ui_sdk_integration.py`
|
||||
|
||||
### Gaps to close — DONE
|
||||
|
||||
- ~~`src/basic_memory/mcp/tools/ui_sdk.py` branch coverage~~ — `tests/mcp/test_ui_sdk.py`
|
||||
- ~~`src/basic_memory/mcp/ui/sdk.py` and `src/basic_memory/mcp/ui/templates.py` branch coverage~~ — `tests/mcp/test_ui_templates.py` + `tests/mcp/test_ui_resources.py`
|
||||
|
||||
### Planned additions — DONE
|
||||
|
||||
- ~~Add unit tests for UI SDK metadata generation and template selection branches.~~ **DONE** — 31 tests
|
||||
- ~~Add integration assertion for variant-specific resource URIs and metadata payload shape.~~ **DONE**
|
||||
|
||||
### 6) Watch Command (`8df88e4`) — DONE
|
||||
|
||||
### Acceptance criteria
|
||||
|
||||
- `basic-memory watch` starts and processes create/update/delete events.
|
||||
- Watch restart/reload path does not duplicate watchers.
|
||||
- Cloud-mode projects are excluded from active watcher set.
|
||||
|
||||
### Existing coverage anchor points
|
||||
|
||||
- `tests/cli/test_watch.py`
|
||||
- `tests/sync/test_coordinator.py`
|
||||
- `tests/sync/test_watch_service_reload.py`
|
||||
|
||||
### Planned additions — DONE
|
||||
|
||||
- ~~Add one stress-style integration test for rapid file changes and watcher stability.~~ **DONE** — `tests/sync/test_watch_service_stress.py` (3 tests: 50-file batch, mixed add/modify/delete batch, rapid modifications to same file)
|
||||
|
||||
### 7) CLI JSON Output (`a47c9c0`) — DONE
|
||||
|
||||
### Acceptance criteria
|
||||
|
||||
- `--format json` returns valid JSON with stable keys for success paths.
|
||||
- Error paths also return JSON-shaped output with correct non-zero exits.
|
||||
- Default human output remains unchanged.
|
||||
|
||||
### Existing coverage anchor points
|
||||
|
||||
- `tests/cli/test_cli_tool_json_output.py`
|
||||
- `test-int/cli/test_cli_tool_json_integration.py`
|
||||
|
||||
### Planned additions — DONE
|
||||
|
||||
- ~~Add one failure-path integration test per high-use tool command.~~ **DONE** — `test-int/cli/test_cli_tool_json_failure_integration.py` (4 tests: read-note not found, write-note missing content, write→read roundtrip, recent-activity empty project)
|
||||
|
||||
### 8) Search/Edit and Metadata Fixes (`530cbac`, `f1d50c2`, `8838571`, `009e849`) — DONE
|
||||
|
||||
### Acceptance criteria
|
||||
|
||||
- Metadata filters produce consistent results on SQLite and Postgres.
|
||||
- `tag:` shorthand works alone and with mixed query terms.
|
||||
- Fast write/edit paths preserve `external_id` and metadata integrity.
|
||||
|
||||
### Existing coverage anchor points
|
||||
|
||||
- `tests/repository/test_metadata_filters.py`
|
||||
- `tests/repository/test_search_repository.py`
|
||||
- `tests/services/test_search_service.py`
|
||||
|
||||
### Planned additions — DONE
|
||||
|
||||
- ~~Add Postgres-specific metadata filter edge-case tests to mirror SQLite assertions exactly.~~ **DONE** — `tests/repository/test_metadata_filters_edge_cases.py` (6 tests: missing field, AND logic, contains single-element array, nested path missing intermediate, $gte/$lte boundaries, $between inclusive — all pass on both SQLite and Postgres)
|
||||
|
||||
### 9) Compatibility and Hotfix Regression Pack (`c46d7a6`, `a0e754b`, `343a6e1`, `24ca5f6`, `e3ced49`, `8489a3d`, `b609c4e`, `f6e0a5b`, `7624a20`)
|
||||
|
||||
### Acceptance criteria
|
||||
|
||||
- Legacy endpoints required by older CLI versions function without `405` (`GET /projects/projects`, `POST /projects/projects`, `POST /projects/config/sync`).
|
||||
- Entity creation conflicts map to conflict status (not 500).
|
||||
- `recent_activity` prompt defaults are correct.
|
||||
- No spurious `metadata: {}` in serialized frontmatter.
|
||||
- Tigris/rclone uses global consistency headers for all transaction types.
|
||||
- `bm --version` fast path avoids heavy import path and remains responsive.
|
||||
- Default SQLite DB path is isolated by config dir.
|
||||
|
||||
### Gaps to close
|
||||
|
||||
- ~~Commits with no direct tests added (`c46d7a6`, `344e651`, `f6e0a5b`) need explicit regression tests.~~ **DONE**
|
||||
|
||||
### Planned additions — DONE
|
||||
|
||||
- ~~Add API compat test covering all legacy endpoint methods and payloads.~~ **DONE** — `test_legacy_v1_add_project_endpoint`, `test_legacy_v1_sync_config_endpoint`
|
||||
- ~~Add CLI fast-path test for `--version` import behavior/performance guard.~~ **DONE** — `test_bm_version_does_not_import_heavy_modules`
|
||||
- ~~Add empty metadata serialization regression test.~~ **DONE** — `test_schema_to_markdown_empty_metadata_no_metadata_key`
|
||||
- Add migration safety test for SQLite generated columns (`VIRTUAL` expectation) — deferred, low risk.
|
||||
|
||||
## MCP Manual Verification Plan (LLM Tool Calls)
|
||||
|
||||
Run after automated tests pass.
|
||||
|
||||
### Setup
|
||||
|
||||
- Start MCP server: `basic-memory mcp --transport stdio`
|
||||
- Use an MCP-capable client and issue tool calls directly.
|
||||
|
||||
### Manual scenarios
|
||||
|
||||
- Schema: call `schema_validate`, `schema_infer`, and `schema_diff` on known fixtures.
|
||||
- Schema: verify error and success payloads match acceptance criteria.
|
||||
- Semantic search: call `search_notes` with `search_type=text|vector|hybrid`.
|
||||
- Semantic search: verify ranking relevance on semantic fixture queries.
|
||||
- Routing: call tools with explicit project on mixed local/cloud setup.
|
||||
- Routing: verify success/failure paths with and without API key.
|
||||
- Permalink routing: read/write/search notes across projects with colliding titles.
|
||||
- Permalink routing: verify memory URL routing correctness.
|
||||
- UI/TUI: call `search_notes` and `read_note` with UI variants and `output_format=text|json`.
|
||||
- UI/TUI: verify payload/resource format and metadata completeness.
|
||||
|
||||
## Implementation Backlog (Ordered)
|
||||
|
||||
1. ~~Fill schema MCP/client/router coverage gaps.~~ **DONE** — 18 tests in `test_tool_schema.py` + `test_client_schema.py`
|
||||
2. ~~Fill semantic search MCP + Postgres repository gaps.~~ **DONE** — 20 tests in `test_postgres_search_repository_unit.py` + `test_tool_search.py`
|
||||
3. ~~Add compatibility regression tests (legacy endpoints, migration, version fast path).~~ **DONE** — 5 tests across 3 files (see below)
|
||||
4. ~~Add feature-level integration tests (permalinks, watch, CLI JSON, metadata filters).~~ **DONE** — 15 tests across 4 files (see items 4, 6, 7, 8 above)
|
||||
5. ~~Expand UI SDK and template branch tests.~~ **DONE** — 31 tests in `test_ui_templates.py` + `test_ui_sdk.py` + `test_ui_resources.py`
|
||||
6. ~~Run full gate and capture results in a short release readiness summary.~~ **DONE** — see results below
|
||||
|
||||
### Full Gate Results (`just check`)
|
||||
|
||||
| Phase | Result |
|
||||
|-------|--------|
|
||||
| lint | PASS |
|
||||
| format | PASS |
|
||||
| typecheck | PASS |
|
||||
| Unit tests (SQLite) | 1788 passed, 15 skipped |
|
||||
| Integration tests (SQLite) | 243 passed, 4 skipped, 10 deselected |
|
||||
| Unit tests (Postgres) | 1760 passed, 28 skipped |
|
||||
| Integration tests (Postgres) | 234 passed, 13 skipped, 10 deselected |
|
||||
|
||||
**0 failures. 10 deselected = semantic benchmark tests (run separately via `just test-semantic`).**
|
||||
|
||||
### Item 3 Details — Compatibility Regression Tests
|
||||
|
||||
| Test | File | What it covers |
|
||||
|------|------|----------------|
|
||||
| `test_legacy_v1_add_project_endpoint` | `tests/api/v2/test_project_router.py` | POST `/projects/projects` legacy route reachable (idempotent path) |
|
||||
| `test_legacy_v1_sync_config_endpoint` | `tests/api/v2/test_project_router.py` | POST `/projects/config/sync` legacy route reachable |
|
||||
| `test_bm_version_does_not_import_heavy_modules` | `tests/cli/test_cli_exit.py` | `bm --version` fast path does not load `basic_memory.mcp` |
|
||||
| `test_schema_to_markdown_empty_metadata_no_metadata_key` | `tests/markdown/test_entity_parser_error_handling.py` | `schema_to_markdown()` with `entity_metadata={}` emits no `metadata:` key |
|
||||
| `test_legacy_v1_list_projects_endpoint` | `tests/api/v2/test_project_router.py` | (pre-existing) GET `/projects/projects` legacy route |
|
||||
|
||||
**Suite totals after item 3: 1764 passed, 15 skipped, 0 failures.**
|
||||
|
||||
## Suggested Commands
|
||||
|
||||
- Full suite: `just check`
|
||||
- Fast loop: `just fast-check`
|
||||
- E2E consistency: `just doctor`
|
||||
- SQLite focused: `just test-sqlite`
|
||||
- Postgres focused: `just test-postgres`
|
||||
- Schema integration: `pytest test-int/test_schema -q`
|
||||
- Semantic + repo focus: `pytest tests/repository/test_postgres_search_repository.py tests/mcp/test_tool_search.py tests/services/test_semantic_search.py -q`
|
||||
- MCP integration focus: `pytest test-int/mcp -q`
|
||||
|
||||
## Exit Criteria for This Plan
|
||||
|
||||
- All feature acceptance criteria above are validated.
|
||||
- All identified high-priority coverage gaps are addressed or explicitly documented as intentional.
|
||||
- Manual MCP verification scenarios complete with no P0/P1 findings.
|
||||
@@ -1,318 +0,0 @@
|
||||
# v0.19.0 Release Notes
|
||||
|
||||
## Overview
|
||||
|
||||
v0.19.0 is a major release that introduces semantic vector search, a schema validation system,
|
||||
project-prefixed permalinks, per-project cloud routing, and a significant upgrade to FastMCP 3.0.
|
||||
It includes 90+ commits since v0.18.0 spanning new features, architectural improvements, and
|
||||
stability fixes across both SQLite and Postgres backends.
|
||||
|
||||
---
|
||||
|
||||
## Major Features
|
||||
|
||||
### Semantic Vector Search
|
||||
|
||||
Full vector and hybrid search for SQLite (via sqlite-vec) and Postgres (via pgvector).
|
||||
|
||||
- **Hybrid search mode** combines full-text search (FTS) with vector similarity for best results
|
||||
- **Score-based fusion** replaces RRF for hybrid ranking — `max(vec, fts) + 0.3 * min(vec, fts)` preserves dominant signals and rewards dual-source agreement (#577)
|
||||
- **Default search mode** is now `hybrid` when semantic search is enabled, `text` when disabled
|
||||
- Embedding providers: FastEmbed (local, default) or OpenAI API
|
||||
- Configurable similarity threshold via `semantic_min_similarity` (default 0.55)
|
||||
- Per-query `min_similarity` override on `search_notes` tool
|
||||
- Auto-backfill: existing entities get embeddings generated on first startup
|
||||
- Backend-specific distance-to-similarity conversion (cosine for SQLite, inner product for Postgres)
|
||||
- FTS fallback: if semantic dependencies are missing, search gracefully degrades to text-only
|
||||
- sqlite-vec knn `k` parameter capped at 4096 to prevent backend errors
|
||||
|
||||
**Configuration:**
|
||||
```json
|
||||
{
|
||||
"semantic_search_enabled": true,
|
||||
"semantic_embedding_provider": "fastembed",
|
||||
"semantic_embedding_model": "bge-small-en-v1.5",
|
||||
"semantic_min_similarity": 0.55
|
||||
}
|
||||
```
|
||||
|
||||
**Usage:**
|
||||
```
|
||||
search_notes("machine learning concepts", search_type="hybrid")
|
||||
search_notes("similar to my notes on coffee", search_type="vector")
|
||||
search_notes("exact phrase match", search_type="text")
|
||||
search_notes("broad search", min_similarity=0.3) # lower threshold for more results
|
||||
```
|
||||
|
||||
### Schema System
|
||||
|
||||
Validate note structure against user-defined schemas with frontmatter-based rules.
|
||||
|
||||
- Define schemas as YAML in note frontmatter with field types, required fields, and constraints
|
||||
- Frontmatter validation during sync — malformed notes get clear error messages
|
||||
- Schema inference from existing notes to bootstrap schemas from your content
|
||||
- Schema diff to compare two schemas and see changes
|
||||
- Available via MCP tools and CLI
|
||||
|
||||
### Project-Prefixed Permalinks
|
||||
|
||||
Permalinks now include the project name for unambiguous cross-project references.
|
||||
|
||||
- Memory URLs like `memory://project-name/folder/note` route to the correct project
|
||||
- Existing non-prefixed permalinks continue to work (backwards compatible)
|
||||
- Controlled by `permalinks_include_project` config (default: true)
|
||||
- `build_context` and `search_notes` auto-detect project from URL prefix
|
||||
|
||||
### Per-Project Cloud Routing
|
||||
|
||||
Individual projects can be routed through the cloud while others stay local.
|
||||
|
||||
- Set a project to cloud mode: `bm project set-cloud research`
|
||||
- Revert to local: `bm project set-local research`
|
||||
- Uses API key authentication: `bm cloud set-key bmc_abc123...`
|
||||
- MCP tools automatically route based on each project's mode
|
||||
- Local MCP server (`bm mcp`) still uses local routing for all projects by default
|
||||
- `--local` and `--cloud` CLI flags override per-command
|
||||
|
||||
### Workspace Selection
|
||||
|
||||
Cloud projects can target specific workspaces for multi-tenant environments.
|
||||
|
||||
- `workspace` parameter on MCP tools for explicit workspace targeting
|
||||
- CLI workspace-aware project listing with `bm project list`
|
||||
- Spinner feedback while fetching cloud projects
|
||||
|
||||
---
|
||||
|
||||
## New Tools and Capabilities
|
||||
|
||||
### Dashboard (`bm project info`)
|
||||
|
||||
`bm project info` now displays an htop-inspired compact dashboard with:
|
||||
|
||||
- Horizontal bar charts for note types (top 5)
|
||||
- Embedding coverage bar with Unicode block characters
|
||||
- Colored status dots for at-a-glance health
|
||||
- `EmbeddingStatus` schema and `get_embedding_status()` service method for programmatic access
|
||||
|
||||
### Unified Metadata Search
|
||||
|
||||
`search_by_metadata` has been merged into `search_notes` — one tool for all searches.
|
||||
`query` is now optional, so you can search purely by frontmatter metadata.
|
||||
|
||||
```
|
||||
search_notes(metadata_filters={"status": "in-progress"})
|
||||
search_notes(metadata_filters={"tags": ["security", "oauth"]})
|
||||
search_notes(metadata_filters={"priority": {"$in": ["high", "critical"]}})
|
||||
search_notes(metadata_filters={"schema.confidence": {"$gt": 0.7}})
|
||||
search_notes(tags=["security"]) # convenience shorthand
|
||||
search_notes(status="draft") # convenience shorthand
|
||||
```
|
||||
|
||||
### JSON Output Mode
|
||||
|
||||
All MCP tools now support `output_format="json"` for machine-readable responses.
|
||||
|
||||
- Default remains `"text"` for human-readable output (no breaking changes)
|
||||
- `build_context` defaults to `"json"` with slimmed payloads (redundant fields stripped)
|
||||
- CLI tool commands support `--format json` flag
|
||||
|
||||
### `tag:` Search Shorthand
|
||||
|
||||
Search by tag using convenient shorthand syntax.
|
||||
|
||||
```
|
||||
search_notes("tag:security")
|
||||
search_notes("tag:coffee AND tag:brewing")
|
||||
```
|
||||
|
||||
### Entity User Tracking
|
||||
|
||||
Entities now track `created_by` and `last_updated_by` fields for attribution.
|
||||
|
||||
### Improved Search Result Content (#609)
|
||||
|
||||
Search results now surface more relevant context:
|
||||
|
||||
- `matched_chunk_text` populated for FTS-only hybrid results (no more fallback to truncated content)
|
||||
- `TOP_CHUNKS_PER_RESULT` increased from 3 to 5, catching answers deeper in large notes (~2700 → ~4500 chars)
|
||||
- `CONTENT_DISPLAY_LIMIT` doubled from 2000 to 4000 chars for results without matched chunks
|
||||
|
||||
### `write_note` Overwrite Guard (#632)
|
||||
|
||||
`write_note` is now non-idempotent by default. If a note already exists, the tool returns an
|
||||
error instead of silently overwriting. Pass `overwrite=True` to replace, or use `edit_note`
|
||||
for incremental updates. Config option `write_note_overwrite_default` restores the old upsert
|
||||
behavior.
|
||||
|
||||
---
|
||||
|
||||
## Architecture Changes
|
||||
|
||||
### Score-Based Hybrid Fusion (#577)
|
||||
|
||||
RRF (Reciprocal Rank Fusion) compressed all fused scores to ~0.016, destroying ranking
|
||||
differentiation. The new formula `max(vec, fts) + FUSION_BONUS * min(vec, fts)` preserves
|
||||
dominant signals and rewards dual-source agreement. Zero-score results now produce zero
|
||||
fused score instead of receiving a 0.1 weight floor.
|
||||
|
||||
### FastMCP 3.0 Upgrade
|
||||
|
||||
Upgraded from FastMCP 2.12.3 to 3.0.1.
|
||||
|
||||
- Tool annotations (`readOnlyHint`, `openWorldHint`) for better client integration
|
||||
- Improved MCP protocol compliance
|
||||
- Better error handling and context management
|
||||
|
||||
### Prompts Call MCP Tools Directly
|
||||
|
||||
MCP prompts (`search`, `continue_conversation`) now call MCP tools directly instead of
|
||||
going through API endpoints. This fixes empty results in discovery mode and ensures prompts
|
||||
use the same resolution logic as tools (including LinkResolver fallback).
|
||||
|
||||
### build_context LinkResolver Fallback
|
||||
|
||||
`build_context` now falls back to LinkResolver when an exact permalink lookup returns empty.
|
||||
This uses the same 7-strategy resolution pipeline as `read_note`, so callers no longer get
|
||||
empty results for valid note identifiers that don't match exact permalinks.
|
||||
|
||||
### Sync Handles Semantic Dependency Errors Gracefully
|
||||
|
||||
When sqlite-vec or another embedding provider is unavailable, `sync_file` now catches
|
||||
`SemanticDependenciesMissingError` separately. The entity is created and FTS-indexed
|
||||
successfully — only vector embeddings are skipped, with a clear warning:
|
||||
|
||||
```
|
||||
WARNING: Semantic search dependencies missing — vector embeddings skipped for path=note.md.
|
||||
Run 'bm reindex --embeddings' after resolving the dependency issue.
|
||||
```
|
||||
|
||||
### Unified Project Path
|
||||
|
||||
Cloud projects with bisync now store the local filesystem path in `path` (not the Docker
|
||||
container path). Config migration automatically promotes `local_sync_path` → `path` for
|
||||
existing configs.
|
||||
|
||||
---
|
||||
|
||||
## CLI Improvements
|
||||
|
||||
### Status and Doctor Default to Local Routing
|
||||
|
||||
`bm status` and `bm doctor` now default to local routing since they scan the local filesystem.
|
||||
Previously, cloud-mode projects would route these commands to the cloud API, which returned
|
||||
Docker-internal paths that don't exist locally.
|
||||
|
||||
### `--format json` for CLI Tool Commands
|
||||
|
||||
All `bm tool` subcommands support `--format json` for machine-readable output, enabling
|
||||
integration with scripts and plugins.
|
||||
|
||||
### `--json` for Top-Level CLI Commands
|
||||
|
||||
Five additional CLI commands now support `--json` for machine-readable output:
|
||||
|
||||
- `bm status --json` — sync report with new/modified/deleted/moved files and skipped files
|
||||
- `bm project list --json` — structured project list with name, paths, routing mode, and defaults
|
||||
- `bm schema validate --json` — validation report with per-note pass/fail, warnings, and errors
|
||||
- `bm schema infer --json` — field frequency analysis and suggested schema definition
|
||||
- `bm schema diff --json` — drift report with new fields, dropped fields, and cardinality changes
|
||||
|
||||
This complements the existing `bm project info --json` and `bm tool --format json` support,
|
||||
making all major CLI commands scriptable for CI pipelines and automation.
|
||||
|
||||
### Cloud Promo and Analytics
|
||||
|
||||
- Cloud promo panel shown on first run or version bump with OSS discount code
|
||||
- Anonymous usage telemetry via Umami Cloud (promo/login funnel events only)
|
||||
- Opt out with `BASIC_MEMORY_NO_PROMOS=1`
|
||||
- No PII, no file contents, no per-command tracking
|
||||
- See [Telemetry](https://github.com/basicmachines-co/basic-memory#telemetry) in README
|
||||
|
||||
---
|
||||
|
||||
## Bug Fixes
|
||||
|
||||
- **#577**: RRF fusion compressed all hybrid scores to ~0.016, destroying ranking differentiation
|
||||
- **#582**: build_context returns empty results on valid note identifiers
|
||||
- **#575**: Remove hardcoded "main" default from default_project
|
||||
- **#595**: recent_activity dedup and pagination across MCP tools
|
||||
- **#593**: Backend-specific distance-to-similarity conversion
|
||||
- **#592**: Strip NUL bytes from content before PostgreSQL search indexing
|
||||
- **#562**: Use VIRTUAL instead of STORED columns in SQLite migration
|
||||
- **#558**: Add X-Tigris-Consistent headers to all rclone commands
|
||||
- **#541**: Handle EntityCreationError as conflict
|
||||
- **#536**: Stabilize metadata filters on Postgres
|
||||
- **#533**: Fix recent_activity prompt defaults
|
||||
- **#530**: Prevent spurious `metadata: {}` in frontmatter output
|
||||
- **#601**: Return matched chunk text in search results
|
||||
- **#606**: Accept `null` for `expected_replacements` in `edit_note`
|
||||
- **#579, #607**: Guard against closed streams in promo panel and missing vector tables on shutdown
|
||||
- **#609**: FTS-only hybrid results missing `matched_chunk_text`; content limits too conservative
|
||||
- **#631**: `build_context` related_results schema validation failure — replaced fragile `_slim_context()` stripping with Pydantic `exclude=True` field config
|
||||
- **#630**: Skip workspace resolution when client factory is active — prevents 401 errors in cloud MCP server mode
|
||||
- **#30**: `tag:` prefix query fails with hybrid search — moved tag prefix parsing to MCP tool level so it works with all search modes
|
||||
- **#31**: `search_notes` returns cluttered observation/relation-level results — now defaults to entity-level results
|
||||
- **#28**: `schema_infer` and `schema_diff` return raw Pydantic models as "undefined" in LLM output — added markdown formatters
|
||||
- Fix `schema_validate` identifier resolution (now uses LinkResolver) and text rendering (markdown formatter)
|
||||
- **#634**: `schema_validate` and `schema_diff` use stale database metadata instead of reading schema definitions from file — now reads frontmatter directly from the file with fallback to database metadata
|
||||
- Fix `Post(**metadata)` crash when frontmatter contains `content` or `handler` keys
|
||||
- Fix list-valued frontmatter fields (`title`, `type`) crashing on `.strip()` — now coerced to strings
|
||||
- Cap sqlite-vec knn `k` parameter at 4096 to prevent backend errors
|
||||
- Parameterize SQL queries in search repository type filters
|
||||
- Double-default display in project list
|
||||
- `ensure_frontmatter_on_sync` default changed to `True`
|
||||
- Status/doctor commands fail with cloud-mode projects (Docker path error)
|
||||
- Prompts return "0 projects" in discovery mode
|
||||
|
||||
---
|
||||
|
||||
## Security
|
||||
|
||||
- Upgrade `cryptography` for CVE advisory
|
||||
- Upgrade `python-multipart` for security advisory
|
||||
|
||||
---
|
||||
|
||||
## Internal / Developer
|
||||
|
||||
- **#598**: Upgrade FastMCP 2.12.3 → 3.0.1 with tool annotations
|
||||
- **#594**: Add `ty` as supplemental type checker
|
||||
- **#538**: Add fast feedback loop tooling (`just fast-check`, `just doctor`, `just testmon`)
|
||||
- **#600**: Rename `entity_type` to `note_type` for consistency
|
||||
- **#596**: Fix CLI runtime defects and audit regressions
|
||||
- CLI refactoring and workspace-aware cloud project listing
|
||||
- Split and speed up PR test matrix in CI
|
||||
- Fix CI: collect coverage from test jobs instead of re-running all tests
|
||||
- Create `search_vector_chunks` in test fixtures for Postgres compatibility
|
||||
|
||||
---
|
||||
|
||||
## Configuration Changes
|
||||
|
||||
| Setting | Old Default | New Default | Notes |
|
||||
|---------|-------------|-------------|-------|
|
||||
| `semantic_search_enabled` | `false` | `true` | Semantic search on by default |
|
||||
| `ensure_frontmatter_on_sync` | `false` | `true` | Frontmatter added during sync |
|
||||
| `permalinks_include_project` | `false` | `true` | Project prefix in permalinks |
|
||||
|
||||
---
|
||||
|
||||
## Upgrade Notes
|
||||
|
||||
- **Semantic search dependencies** are now included by default. If sqlite-vec fails to load,
|
||||
search gracefully falls back to FTS. Run `bm reindex --embeddings` to generate embeddings
|
||||
for existing content.
|
||||
- **Hybrid search scoring** has changed from RRF to score-based fusion. Search result ordering
|
||||
may differ — results should be more accurate with better score differentiation.
|
||||
- **`search_by_metadata`** is removed as a standalone tool. Use `search_notes` with
|
||||
`metadata_filters` instead (same parameters, same behavior).
|
||||
- **Project-prefixed permalinks** are enabled by default. Existing notes keep their current
|
||||
permalinks until modified. Set `permalinks_include_project: false` to disable.
|
||||
- **Frontmatter on sync** is now enabled by default. Files without frontmatter will have it
|
||||
added on next sync. Set `ensure_frontmatter_on_sync: false` to preserve old behavior.
|
||||
- **Config migration** runs automatically for cloud projects with bisync — `local_sync_path`
|
||||
is promoted to `path` so filesystem operations work correctly.
|
||||
- **`write_note` is no longer idempotent** — calls to `write_note` for existing notes now
|
||||
return an error unless `overwrite=True` is passed. Use `edit_note` for incremental changes,
|
||||
or set `write_note_overwrite_default: true` in config to restore the old behavior.
|
||||
@@ -1,209 +0,0 @@
|
||||
# Semantic Search Manual Test Log
|
||||
|
||||
## Overview
|
||||
|
||||
Manual test session for semantic (vector) search on the main project.
|
||||
- Date: 2026-02-15
|
||||
- Database: ~/.basic-memory/memory.db (SQLite)
|
||||
- Entities: 456 embedded, 2714 vector chunks
|
||||
- Search index: 2390 FTS entries
|
||||
- Embedding model: default (384-dim, sqlite-vec)
|
||||
|
||||
## Test Plan
|
||||
|
||||
1. **Search Type Routing** — verify vector/hybrid/text dispatch, invalid search_type handling
|
||||
2. **Conceptual Queries** — natural language where vector should beat FTS
|
||||
3. **Keyword Queries** — exact terms where FTS should be strong
|
||||
4. **Hybrid Ranking** — queries where both FTS and vector contribute
|
||||
5. **Result Types** — entities, observations, relations in vector results
|
||||
6. **Filters + Vector** — combine vector with types/entity_types/after_date
|
||||
7. **Edge Cases** — short queries, long queries, empty, special chars, no-match
|
||||
8. **Pagination** — page > 1, page_size respected
|
||||
|
||||
---
|
||||
|
||||
## Test Results
|
||||
|
||||
### Test 1: Search Type Routing
|
||||
|
||||
#### 1a: search_type="semantic" (invalid value)
|
||||
- **Input:** query="how does the knowledge graph work", search_type="semantic"
|
||||
- **Expected:** error or explicit fallback
|
||||
- **Actual:** Silently falls through to text search (else branch in search.py:430)
|
||||
- **Verdict:** BUG — should either be a recognized alias for "vector" or return an error
|
||||
|
||||
#### 1b: search_type="vector"
|
||||
- **Input:** query="keeping AI context between sessions", search_type="vector"
|
||||
- **Actual:** 5 results, scores ~0.58-0.59, found "Maintaining context across conversation boundaries" observation
|
||||
- **Verdict:** PASS
|
||||
|
||||
#### 1c: search_type="text" with conceptual query
|
||||
- **Input:** query="keeping AI context between sessions", search_type="text"
|
||||
- **Actual:** 0 results (no exact keyword match)
|
||||
- **Verdict:** PASS (expected — FTS requires token overlap)
|
||||
|
||||
#### 1d: search_type="hybrid" with conceptual query
|
||||
- **Input:** query="keeping AI context between sessions", search_type="hybrid"
|
||||
- **Actual:** 5 results, same ranking as vector (FTS contributed nothing here)
|
||||
- **Verdict:** PASS
|
||||
|
||||
#### 1e: search_type="text" with keyword query
|
||||
- **Input:** query="OAuth authentication", search_type="text"
|
||||
- **Actual:** 3 results — AUTH.md Supabase OAuth, OAuth Rip-and-Replace, OAuth Integration Analysis
|
||||
- **Verdict:** PASS
|
||||
|
||||
#### 1f: search_type="vector" with keyword query
|
||||
- **Input:** query="OAuth authentication", search_type="vector"
|
||||
- **Actual:** Same top results as text (keyword-rich content also scores well in vector space)
|
||||
- **Verdict:** PASS
|
||||
|
||||
---
|
||||
|
||||
### Test 2: Conceptual Queries (vector advantage)
|
||||
|
||||
#### 2a: Natural language question
|
||||
- **Input:** query="why do AI assistants forget things", search_type="vector"
|
||||
- **Actual:** 5 results — Manual Testing Session, "Balance security and usability" observation, "Tools should match thought patterns" observation. Scores ~0.56-0.57
|
||||
- **Vector advantage:** Found conceptually related content despite no exact keyword overlap
|
||||
- **Verdict:** PASS
|
||||
|
||||
#### 2b: Same query, text search
|
||||
- **Input:** query="why do AI assistants forget things", search_type="text"
|
||||
- **Actual:** 1 result — "What is Basic Memory?" (likely matched on "AI" token)
|
||||
- **Verdict:** PASS (demonstrates vector advantage — text barely matched)
|
||||
|
||||
#### 2c: Domain concept with no jargon
|
||||
- **Input:** query="pricing strategy for cloud product", search_type="vector"
|
||||
- **Actual:** 3 results — SPEC-16 MCP Cloud Service Consolidation, knowledge architecture observation, Visual Knowledge Spaces relation. Scores ~0.56-0.57
|
||||
- **Verdict:** PASS (found cloud-related content conceptually)
|
||||
|
||||
#### 2d: Technical concept, long query
|
||||
- **Input:** query="SQLite performance optimization WAL mode concurrent writes", search_type="vector"
|
||||
- **Actual:** 3 results — SPEC-11 API Performance Optimization, Real-Time Updates with WebSockets, marketing status update. Scores ~0.55-0.58
|
||||
- **Verdict:** PASS (found performance-related content)
|
||||
|
||||
---
|
||||
|
||||
### Test 3: Keyword Queries (FTS strength)
|
||||
|
||||
#### 3a: Exact term match — "OAuth authentication"
|
||||
- **Text:** 3 results with high relevance (exact matches in titles)
|
||||
- **Vector:** Same top results (keyword overlap helps vector too)
|
||||
- **Verdict:** PASS — FTS and vector converge on keyword-rich queries
|
||||
|
||||
#### 3b: "OAuth" single keyword, hybrid mode
|
||||
- **Input:** query="OAuth", search_type="hybrid"
|
||||
- **Actual:** 5 results — Basic Memory Coding Guide, AI Collaboration Examples, SPEC-18, daily note, Manual Testing Session. FTS + vector blended. Scores ~0.016-0.032
|
||||
- **Note:** Top hybrid result is "Basic Memory Coding Guide" not an OAuth-specific doc — suggests hybrid scoring may dilute strong FTS matches
|
||||
- **Verdict:** PASS but hybrid ranking questionable for single-keyword queries
|
||||
|
||||
---
|
||||
|
||||
### Test 4: Hybrid Ranking
|
||||
|
||||
#### 4a: Hybrid vs vector on "OAuth authentication"
|
||||
- **Hybrid with entity_types=["entity"]:** 5 results — RLS Implementation Lessons, Cloud Readiness Assessment, AUTH.md OAuth, Core Service Implementation, OAuth Rip-and-Replace. Scores ~0.016-0.023
|
||||
- **Vector with entity_types=["entity"]:** 5 results — Core Service Implementation, SPEC-13 CLI Auth, Coding Guide, Authentication Service, ADR Production Auth. Scores ~0.55-0.60
|
||||
- **Observation:** Hybrid surfaces different top results than vector-only. Hybrid found RLS and Cloud Readiness docs that vector didn't prioritize. Different ranking is expected from RRF fusion.
|
||||
- **Verdict:** PASS — hybrid produces meaningfully different ranking
|
||||
|
||||
---
|
||||
|
||||
### Test 5: Result Types
|
||||
|
||||
#### 5a: Vector returns all result types
|
||||
- **Input:** query="keeping AI context between sessions", search_type="vector"
|
||||
- **Entities:** SPEC-18 AI Memory Management Tool (type=entity)
|
||||
- **Relations:** Prompt Builder integrates_with (type=relation)
|
||||
- **Observations:** "Translation layer is key" (type=observation), "Maintaining context across conversation boundaries" (type=observation)
|
||||
- **Verdict:** PASS — all three types appear in vector results
|
||||
|
||||
#### 5b: Observations carry metadata
|
||||
- **Observation result:** category="challenge", content="Maintaining context across conversation boundaries", from_entity="research/ai-knowledge-management-research"
|
||||
- **Verdict:** PASS — category, content, from_entity, tags all present
|
||||
|
||||
#### 5c: Relations carry link info
|
||||
- **Relation result:** relation_type="integrates_with", from_entity="development/features/prompt-builder...", to_entity (present but truncated in some)
|
||||
- **Verdict:** PASS — relation metadata present
|
||||
|
||||
---
|
||||
|
||||
### Test 6: Filters + Vector Search
|
||||
|
||||
#### 6a: entity_types=["entity"] with vector
|
||||
- **Input:** query="OAuth authentication", search_type="vector", entity_types=["entity"]
|
||||
- **Actual:** 5 results, all type="entity" (Core Service Implementation, SPEC-13, Coding Guide, Authentication Service, ADR Auth)
|
||||
- **Verdict:** PASS — filter correctly restricts to entities only
|
||||
|
||||
#### 6b: types=["note"] with vector
|
||||
- **Input:** query="OAuth authentication", search_type="vector", types=["note"]
|
||||
- **Actual:** Same 5 results (all have entity_type="note" in metadata)
|
||||
- **Verdict:** PASS — types filter works with vector search
|
||||
|
||||
#### 6c: after_date with vector
|
||||
- **Input:** query="OAuth authentication", search_type="vector", after_date="2025-06-01"
|
||||
- **Actual:** 3 results — Core Service Implementation, Cloud Web App analysis observation, SPEC-13. Filtered out older OAuth docs.
|
||||
- **Verdict:** PASS — date filter applied correctly
|
||||
|
||||
#### 6d: entity_types=["entity"] with hybrid
|
||||
- **Input:** query="OAuth authentication", search_type="hybrid", entity_types=["entity"]
|
||||
- **Actual:** 5 results, all type="entity" — RLS lessons, Cloud Readiness, AUTH.md OAuth, Core Service, OAuth Rip-and-Replace
|
||||
- **Verdict:** PASS — filter works with hybrid mode too
|
||||
|
||||
#### 6e: types=["entity"] with vector (WRONG filter name)
|
||||
- **Input:** query="OAuth authentication", search_type="vector", types=["entity"]
|
||||
- **Actual:** 0 results
|
||||
- **Note:** `types` filters by entity_type metadata (e.g., "note", "person"), NOT by SearchItemType. Using types=["entity"] looks for entity_type="entity" which few/no notes have. This is a UX confusion point — the param names are ambiguous.
|
||||
- **Verdict:** PASS (correct behavior) but USABILITY ISSUE — easy to confuse types vs entity_types
|
||||
|
||||
---
|
||||
|
||||
### Test 7: Edge Cases
|
||||
|
||||
#### 7a: Single character query
|
||||
- **Input:** query="x", search_type="vector"
|
||||
- **Actual:** 3 results — "Self-contained application bundle" observation, Non-Markdown File Support relation, quick-win-tools entity. Scores ~0.57-0.59
|
||||
- **Note:** Single character still produces an embedding and returns results. Quality is low/random as expected.
|
||||
- **Verdict:** PASS (no crash, returns results)
|
||||
|
||||
#### 7b: Whitespace-only query
|
||||
- **Input:** query=" ", search_type="vector"
|
||||
- **Actual:** 0 results
|
||||
- **Verdict:** PASS (handled gracefully — _check_vector_eligible strips and rejects empty)
|
||||
|
||||
#### 7c: Query with no relevant content
|
||||
- **Input:** query="quantum computing blockchain", search_type="vector"
|
||||
- **Actual:** 3 results — Inter-Agent Communication relation, Self-contained bundle observation, JSON-LD interop observation. Scores ~0.54
|
||||
- **Note:** Still returns results because vector search always finds nearest neighbors. Scores are lower (~0.54) than relevant queries (~0.58-0.60). No relevance threshold applied.
|
||||
- **Verdict:** PASS (expected behavior) but NOTE — no relevance cutoff means irrelevant queries always return something
|
||||
|
||||
---
|
||||
|
||||
### Test 8: Pagination
|
||||
|
||||
#### 8a: Vector search page 2
|
||||
- **Input:** query="keeping AI context between sessions", search_type="vector", page=2, page_size=3
|
||||
- **Actual:** 3 results on page 2, current_page=2. Different results from page 1. Top: "Maintaining context across conversation boundaries" observation (score 0.587)
|
||||
- **Note:** Interestingly, page 2 had a higher-scoring result than some page 1 results. This may indicate pagination doesn't sort globally — it might be paginating within a pre-scored set.
|
||||
- **Verdict:** PASS (pagination works) but POSSIBLE ISSUE — result ordering across pages needs investigation
|
||||
|
||||
---
|
||||
|
||||
## Summary
|
||||
|
||||
### Passing Tests: 20/21
|
||||
|
||||
### Bugs Found
|
||||
1. **search_type="semantic" silently falls through** (Test 1a) — Invalid search_type values fall to the `else` branch and default to text search without any warning. Should either alias "semantic" to "vector" or raise an error.
|
||||
|
||||
### Usability Issues
|
||||
2. **types vs entity_types confusion** (Test 6e) — `types` filters by entity_type metadata (note, person, etc.) while `entity_types` filters by SearchItemType (entity, observation, relation). The naming is ambiguous and easy to mix up.
|
||||
3. **No relevance threshold** (Test 7c) — Vector search always returns nearest neighbors even for completely irrelevant queries. Consider adding a minimum score threshold or at least documenting expected score ranges.
|
||||
4. **Hybrid ranking for single keywords** (Test 3b) — Hybrid mode on simple keyword queries produced less intuitive rankings than pure FTS or pure vector. The RRF fusion may dilute strong FTS signals.
|
||||
|
||||
### Observations
|
||||
- Vector search successfully finds conceptually related content that FTS misses entirely
|
||||
- Score ranges: relevant queries ~0.56-0.60, irrelevant queries ~0.54 (narrow spread)
|
||||
- All three result types (entity, observation, relation) appear correctly in vector results
|
||||
- Filters (entity_types, types, after_date) all work correctly with vector and hybrid modes
|
||||
- Pagination works but cross-page ordering may need investigation
|
||||
@@ -1,382 +0,0 @@
|
||||
# Semantic Search
|
||||
|
||||
This guide covers Basic Memory's semantic (vector) search feature, which adds meaning-based retrieval alongside the existing full-text search.
|
||||
|
||||
## Overview
|
||||
|
||||
Basic Memory's search supports both full-text search (FTS) and semantic retrieval. Semantic search adds vector embeddings that capture the *meaning* of your content, enabling:
|
||||
|
||||
- **Paraphrase matching**: Find "authentication flow" when searching for "login process"
|
||||
- **Conceptual queries**: Search for "ways to improve performance" and find notes about caching, indexing, and optimization
|
||||
- **Hybrid retrieval**: Combine the precision of keyword search with the recall of semantic similarity
|
||||
|
||||
Semantic search is enabled by default when semantic dependencies are available at runtime. It works on both SQLite (local) and Postgres (cloud) backends.
|
||||
|
||||
## Installation
|
||||
|
||||
Semantic search dependencies (fastembed, sqlite-vec, openai) are included in the default `basic-memory` install.
|
||||
|
||||
```bash
|
||||
pip install basic-memory
|
||||
```
|
||||
|
||||
You can always override with `BASIC_MEMORY_SEMANTIC_SEARCH_ENABLED=true|false`.
|
||||
|
||||
### Platform Compatibility
|
||||
|
||||
| Platform | FastEmbed (local) | OpenAI (API) |
|
||||
|---|---|---|
|
||||
| macOS ARM64 (Apple Silicon) | Yes | Yes |
|
||||
| macOS x86_64 (Intel Mac) | No — see workaround below | Yes |
|
||||
| Linux x86_64 | Yes | Yes |
|
||||
| Linux ARM64 | Yes | Yes |
|
||||
| Windows x86_64 | Yes | Yes |
|
||||
|
||||
#### Intel Mac Workaround
|
||||
|
||||
The default install includes FastEmbed, which depends on ONNX Runtime. ONNX Runtime dropped Intel Mac (x86_64) wheels starting in v1.24, so install with a compatible ONNX Runtime pin first:
|
||||
|
||||
```bash
|
||||
pip install basic-memory 'onnxruntime<1.24'
|
||||
```
|
||||
|
||||
After installation, Intel Mac users have two runtime options:
|
||||
|
||||
**Option 1: Use OpenAI embeddings (recommended)**
|
||||
|
||||
```bash
|
||||
export BASIC_MEMORY_SEMANTIC_SEARCH_ENABLED=true
|
||||
export BASIC_MEMORY_SEMANTIC_EMBEDDING_PROVIDER=openai
|
||||
export OPENAI_API_KEY=sk-...
|
||||
```
|
||||
|
||||
**Option 2: Use FastEmbed locally**
|
||||
|
||||
Keep the same pinned installation and use FastEmbed (default provider):
|
||||
|
||||
```bash
|
||||
export BASIC_MEMORY_SEMANTIC_SEARCH_ENABLED=true
|
||||
export BASIC_MEMORY_SEMANTIC_EMBEDDING_PROVIDER=fastembed
|
||||
```
|
||||
|
||||
## Quick Start
|
||||
|
||||
1. Install Basic Memory:
|
||||
|
||||
```bash
|
||||
pip install basic-memory
|
||||
```
|
||||
|
||||
2. (Optional) Explicitly enable semantic search:
|
||||
|
||||
```bash
|
||||
export BASIC_MEMORY_SEMANTIC_SEARCH_ENABLED=true
|
||||
```
|
||||
|
||||
3. Build vector embeddings for your existing content:
|
||||
|
||||
```bash
|
||||
bm reindex --embeddings
|
||||
```
|
||||
|
||||
4. Search using semantic modes:
|
||||
|
||||
```python
|
||||
# Pure vector similarity
|
||||
search_notes("login process", search_type="vector")
|
||||
|
||||
# Hybrid: combines FTS precision with vector recall (recommended)
|
||||
search_notes("login process", search_type="hybrid")
|
||||
|
||||
# Explicit full-text search
|
||||
search_notes("login process", search_type="text")
|
||||
```
|
||||
|
||||
## Configuration Reference
|
||||
|
||||
All settings are fields on `BasicMemoryConfig` and can be set via environment variables (prefixed with `BASIC_MEMORY_`).
|
||||
|
||||
| Config Field | Env Var | Default | Description |
|
||||
|---|---|---|---|
|
||||
| `semantic_search_enabled` | `BASIC_MEMORY_SEMANTIC_SEARCH_ENABLED` | Auto (`true` when semantic deps are available) | Enable semantic search. Required before vector/hybrid modes work. |
|
||||
| `semantic_embedding_provider` | `BASIC_MEMORY_SEMANTIC_EMBEDDING_PROVIDER` | `"fastembed"` | Embedding provider: `"fastembed"` (local), `"openai"` (API), or `"litellm"` (multi-provider API, **experimental** — advanced users only). |
|
||||
| `semantic_embedding_model` | `BASIC_MEMORY_SEMANTIC_EMBEDDING_MODEL` | `"bge-small-en-v1.5"` | Model identifier. Auto-adjusted per provider if left at default. |
|
||||
| `semantic_embedding_dimensions` | `BASIC_MEMORY_SEMANTIC_EMBEDDING_DIMENSIONS` | Provider default | Vector dimensions. 384 for FastEmbed, 1536 for OpenAI/LiteLLM OpenAI. Required when using a non-default LiteLLM model. |
|
||||
| `semantic_embedding_forward_dimensions` | `BASIC_MEMORY_SEMANTIC_EMBEDDING_FORWARD_DIMENSIONS` | Auto | LiteLLM-only override for whether configured dimensions are sent as a provider-side output-size request. |
|
||||
| `semantic_embedding_batch_size` | `BASIC_MEMORY_SEMANTIC_EMBEDDING_BATCH_SIZE` | `2` | Number of texts to embed per batch. |
|
||||
| `semantic_embedding_document_input_type` | `BASIC_MEMORY_SEMANTIC_EMBEDDING_DOCUMENT_INPUT_TYPE` | Auto for known LiteLLM models | Optional LiteLLM `input_type` for indexed document/passages. |
|
||||
| `semantic_embedding_query_input_type` | `BASIC_MEMORY_SEMANTIC_EMBEDDING_QUERY_INPUT_TYPE` | Auto for known LiteLLM models | Optional LiteLLM `input_type` for search queries. |
|
||||
| `semantic_vector_k` | `BASIC_MEMORY_SEMANTIC_VECTOR_K` | `100` | Candidate count for vector nearest-neighbour retrieval. Higher values improve recall at the cost of latency. |
|
||||
|
||||
## Embedding Providers
|
||||
|
||||
### FastEmbed (default)
|
||||
|
||||
FastEmbed runs entirely locally using ONNX models — no API key, no network calls, no cost.
|
||||
|
||||
- **Model**: `BAAI/bge-small-en-v1.5`
|
||||
- **Dimensions**: 384
|
||||
- **Tradeoff**: Smaller model, fast inference, good quality for most use cases
|
||||
|
||||
```bash
|
||||
# Install basic-memory and enable semantic search
|
||||
pip install basic-memory
|
||||
export BASIC_MEMORY_SEMANTIC_SEARCH_ENABLED=true
|
||||
```
|
||||
|
||||
### OpenAI
|
||||
|
||||
Uses OpenAI's embeddings API for higher-dimensional vectors. Requires an API key.
|
||||
|
||||
- **Model**: `text-embedding-3-small`
|
||||
- **Dimensions**: 1536
|
||||
- **Tradeoff**: Higher quality embeddings, requires API calls and an OpenAI key
|
||||
|
||||
```bash
|
||||
export BASIC_MEMORY_SEMANTIC_SEARCH_ENABLED=true
|
||||
export BASIC_MEMORY_SEMANTIC_EMBEDDING_PROVIDER=openai
|
||||
export OPENAI_API_KEY=sk-...
|
||||
```
|
||||
|
||||
### LiteLLM
|
||||
|
||||
> **Experimental — advanced users only.** The LiteLLM provider is experimental and aimed at users comfortable operating remote embedding backends: paid API calls, per-model dimension and input-role configuration, and slower reindexing of large corpora. For most users, FastEmbed (local, default) is recommended. See [LiteLLM Provider](litellm-provider.md) for the caveats and tuning.
|
||||
|
||||
Uses the LiteLLM SDK to call embedding models from providers such as OpenAI, Cohere, Azure, Bedrock, NVIDIA NIM, and other LiteLLM-supported backends. Requires the provider's API credentials.
|
||||
For the full option reference, provider setup examples, and live validation harness, see [LiteLLM Provider](litellm-provider.md).
|
||||
|
||||
```bash
|
||||
export BASIC_MEMORY_SEMANTIC_SEARCH_ENABLED=true
|
||||
export BASIC_MEMORY_SEMANTIC_EMBEDDING_PROVIDER=litellm
|
||||
export BASIC_MEMORY_SEMANTIC_EMBEDDING_MODEL=cohere/embed-english-v3.0
|
||||
export BASIC_MEMORY_SEMANTIC_EMBEDDING_DIMENSIONS=1024
|
||||
export COHERE_API_KEY=...
|
||||
```
|
||||
|
||||
Basic Memory creates vector tables before the first embedding call, so non-default LiteLLM models must set `BASIC_MEMORY_SEMANTIC_EMBEDDING_DIMENSIONS`. The LiteLLM OpenAI default (`openai/text-embedding-3-small`) uses 1536 dimensions automatically.
|
||||
|
||||
For fixed-size LiteLLM models, dimensions are used as Basic Memory's local vector schema and
|
||||
validation size. Basic Memory automatically sends dimensions as a provider-side output-size
|
||||
request for `text-embedding-3` model strings, where LiteLLM/OpenAI support reduced output
|
||||
dimensions. If an Azure/OpenAI deployment uses an arbitrary LiteLLM model string such as
|
||||
`azure/<deployment-name>` and the underlying model supports reduced dimensions, set
|
||||
`BASIC_MEMORY_SEMANTIC_EMBEDDING_FORWARD_DIMENSIONS=true`.
|
||||
|
||||
Some retrieval models are asymmetric: indexed passages and search queries must be embedded with different provider parameters. Basic Memory automatically sets LiteLLM `input_type` for known asymmetric model families:
|
||||
|
||||
- Cohere v3: documents use `search_document`, queries use `search_query`
|
||||
- NVIDIA NIM retrieval models: documents use `passage`, queries use `query`
|
||||
|
||||
For other asymmetric LiteLLM models, set the input types explicitly:
|
||||
|
||||
```bash
|
||||
export BASIC_MEMORY_SEMANTIC_EMBEDDING_DOCUMENT_INPUT_TYPE=passage
|
||||
export BASIC_MEMORY_SEMANTIC_EMBEDDING_QUERY_INPUT_TYPE=query
|
||||
```
|
||||
|
||||
#### Live LiteLLM Validation
|
||||
|
||||
Provider APIs differ in subtle ways: some accept `dimensions`, some require separate
|
||||
document/query roles, and some route through deployment aliases that do not reveal the
|
||||
underlying model name. Before adding or changing LiteLLM model support, run the opt-in live
|
||||
evaluation harness:
|
||||
|
||||
```bash
|
||||
export OPENAI_API_KEY=sk-...
|
||||
export COHERE_API_KEY=...
|
||||
just test-litellm-live
|
||||
```
|
||||
|
||||
The built-in live cases cover:
|
||||
|
||||
| Case | Required key | What it validates |
|
||||
|---|---|---|
|
||||
| `openai/text-embedding-3-small` | `OPENAI_API_KEY` | Standard LiteLLM OpenAI embedding calls and normalized 1536-dimensional output. |
|
||||
| `cohere/embed-english-v3.0` | `COHERE_API_KEY` | Cohere v3 asymmetric `search_document` / `search_query` handling and fixed 1024-dimensional output. |
|
||||
|
||||
The harness embeds two documents and one query, checks vector dimensions and normalization,
|
||||
then verifies the authentication query ranks the authentication document above the distractor.
|
||||
It prints a table with per-model scores, norms, latency, role settings, and dimension-forwarding
|
||||
mode.
|
||||
|
||||
To validate provider aliases or additional LiteLLM backends, save custom JSON cases:
|
||||
|
||||
```bash
|
||||
export AZURE_API_KEY=...
|
||||
export AZURE_API_BASE=https://example.openai.azure.com
|
||||
export AZURE_API_VERSION=2024-02-01
|
||||
|
||||
cat > /tmp/litellm-azure-cases.json <<'JSON'
|
||||
[
|
||||
{
|
||||
"name": "azure-text-embedding-3-small-512",
|
||||
"model": "azure/<deployment-name>",
|
||||
"dimensions": 512,
|
||||
"api_key_env": "AZURE_API_KEY",
|
||||
"forward_dimensions": true
|
||||
}
|
||||
]
|
||||
JSON
|
||||
|
||||
just test-litellm-live --cases-file /tmp/litellm-azure-cases.json
|
||||
```
|
||||
|
||||
NVIDIA NIM retrieval models can be checked the same way:
|
||||
|
||||
```bash
|
||||
export NVIDIA_NIM_API_KEY=...
|
||||
|
||||
cat > /tmp/litellm-nvidia-cases.json <<'JSON'
|
||||
[
|
||||
{
|
||||
"name": "nvidia-embed-qa-4",
|
||||
"model": "nvidia_nim/nvidia/embed-qa-4",
|
||||
"dimensions": 1024,
|
||||
"api_key_env": "NVIDIA_NIM_API_KEY",
|
||||
"document_input_type": "passage",
|
||||
"query_input_type": "query"
|
||||
}
|
||||
]
|
||||
JSON
|
||||
|
||||
just test-litellm-live --cases-file /tmp/litellm-nvidia-cases.json
|
||||
```
|
||||
|
||||
For repeatable local runs, put the same JSON array in a file and pass
|
||||
`just test-litellm-live --cases-file path/to/litellm-cases.json`.
|
||||
|
||||
When switching providers, models, dimensions, or LiteLLM document/query input types, rebuild embeddings:
|
||||
|
||||
```bash
|
||||
bm reindex --embeddings
|
||||
```
|
||||
|
||||
## Search Modes
|
||||
|
||||
### `text` (default)
|
||||
|
||||
Full-text keyword search using FTS5 (SQLite) or tsvector (Postgres). Supports boolean operators (`AND`, `OR`, `NOT`), phrase matching, and prefix wildcards.
|
||||
|
||||
```python
|
||||
search_notes("project AND planning", search_type="text")
|
||||
```
|
||||
|
||||
This is the existing default and does not require semantic search to be enabled.
|
||||
|
||||
### `vector`
|
||||
|
||||
Pure semantic similarity search. Embeds your query and finds the nearest content vectors. Good for conceptual or paraphrase queries where exact keywords may not appear in the content.
|
||||
|
||||
```python
|
||||
search_notes("how to speed up the app", search_type="vector")
|
||||
```
|
||||
|
||||
Returns results ranked by cosine similarity. Individual observations and relations surface as first-class results, not collapsed into parent entities.
|
||||
|
||||
### `hybrid`
|
||||
|
||||
Combines FTS and vector results using score-based fusion. This is generally the best mode when you want both keyword precision and semantic recall.
|
||||
|
||||
```python
|
||||
search_notes("authentication security", search_type="hybrid")
|
||||
```
|
||||
|
||||
Score-based fusion uses the formula `max(vec, fts) + bonus * min(vec, fts)` to preserve the dominant signal while rewarding results found by both methods.
|
||||
|
||||
### When to Use Which
|
||||
|
||||
| Mode | Best For |
|
||||
|---|---|
|
||||
| `text` | Exact keyword matching, boolean queries, tag/category searches |
|
||||
| `vector` | Conceptual queries, paraphrase matching, exploratory searches |
|
||||
| `hybrid` | General-purpose search combining precision and recall |
|
||||
|
||||
## The Reindex Command
|
||||
|
||||
The `bm reindex` command rebuilds search indexes without dropping the database.
|
||||
|
||||
```bash
|
||||
# Rebuild everything (FTS + embeddings if semantic is enabled)
|
||||
bm reindex
|
||||
|
||||
# Only rebuild vector embeddings
|
||||
bm reindex --embeddings
|
||||
|
||||
# Only rebuild the full-text search index
|
||||
bm reindex --search
|
||||
|
||||
# Target a specific project
|
||||
bm reindex -p my-project
|
||||
```
|
||||
|
||||
### When You Need to Reindex
|
||||
|
||||
- **Upgrade note**: Migration now performs a one-time automatic embedding backfill on upgrade.
|
||||
- **Manual enable case**: If you explicitly had `semantic_search_enabled=false` and then turn it on
|
||||
- **Provider change**: After switching between `fastembed`, `openai`, and `litellm`
|
||||
- **Model change**: After changing `semantic_embedding_model`
|
||||
- **Dimension change**: After changing `semantic_embedding_dimensions`
|
||||
- **LiteLLM role change**: After changing `semantic_embedding_document_input_type` or `semantic_embedding_query_input_type`
|
||||
|
||||
The reindex command shows progress with embedded/skipped/error counts:
|
||||
|
||||
```
|
||||
Project: main
|
||||
Building vector embeddings...
|
||||
✓ Embeddings complete: 142 entities embedded, 0 skipped, 0 errors
|
||||
|
||||
Reindex complete!
|
||||
```
|
||||
|
||||
## How It Works
|
||||
|
||||
### Chunking
|
||||
|
||||
Each entity in the search index is split into semantic chunks before embedding:
|
||||
|
||||
- **Headers**: Markdown headers (`#`, `##`, etc.) start new chunks
|
||||
- **Bullets**: Each bullet item (`-`, `*`) becomes its own chunk for granular fact retrieval
|
||||
- **Prose sections**: Non-bullet text is merged up to ~900 characters per chunk
|
||||
- **Long sections**: Oversized content is split with ~120 character overlap to preserve context at boundaries
|
||||
|
||||
Each search index item type (entity, observation, relation) is chunked independently, so observations and relations are embeddable as discrete facts.
|
||||
|
||||
### Deduplication
|
||||
|
||||
Each chunk has a `source_hash` (SHA-256 of the chunk text). On re-sync, unchanged chunks skip re-embedding entirely. This makes incremental updates fast — only modified content triggers API calls or model inference.
|
||||
|
||||
### Hybrid Fusion
|
||||
|
||||
Hybrid search uses score-based fusion to merge FTS and vector results:
|
||||
|
||||
1. Run FTS search to get keyword-ranked results; normalize scores to [0, 1]
|
||||
2. Run vector search to get similarity-ranked results (already [0, 1])
|
||||
3. For each result, compute: `fused = max(vec_score, fts_score) + 0.3 * min(vec_score, fts_score)`
|
||||
4. Sort by fused score
|
||||
|
||||
The dominant signal (whichever source scored higher) is preserved, and dual-source agreement adds a bonus. Unlike rank-based fusion, this approach retains score magnitude — a strong vector match stays strong even without an FTS hit.
|
||||
|
||||
### Observation-Level Results
|
||||
|
||||
Vector and hybrid modes return individual observations and relations as first-class search results, not just parent entities. This means a search for "water temperature for brewing" can surface the specific observation about 205°F without returning the entire "Coffee Brewing Methods" entity.
|
||||
|
||||
## Database Backends
|
||||
|
||||
### SQLite (local)
|
||||
|
||||
- **Vector storage**: [sqlite-vec](https://github.com/asg017/sqlite-vec) virtual table
|
||||
- **Table creation**: At runtime when semantic search is first used — no migration needed
|
||||
- **Embedding table**: `search_vector_embeddings` using `vec0(embedding float[N])` where N is the configured dimensions
|
||||
- **Chunk metadata**: `search_vector_chunks` table stores chunk text, keys, and source hashes
|
||||
|
||||
The sqlite-vec extension is loaded per-connection. Vector tables are created lazily on first use.
|
||||
|
||||
### Postgres (cloud)
|
||||
|
||||
- **Vector storage**: [pgvector](https://github.com/pgvector/pgvector) with HNSW indexing
|
||||
- **Local Docker**: use `docker-compose-postgres.yml` (`pgvector/pgvector:pg17`). Plain `postgres:17` lacks the extension; run `CREATE EXTENSION IF NOT EXISTS vector;` on any external instance before first migration.
|
||||
- **Chunk metadata table**: Created via Alembic migration (`search_vector_chunks` with `BIGSERIAL` primary key)
|
||||
- **Embedding table**: `search_vector_embeddings` created at runtime (dimension-dependent, same pattern as SQLite)
|
||||
- **Index**: HNSW index on the embedding column for fast approximate nearest-neighbour queries
|
||||
|
||||
The Alembic migration creates the dimension-independent chunks table. The embeddings table and HNSW index are deferred to runtime because they depend on the configured vector dimensions.
|
||||
@@ -1,225 +0,0 @@
|
||||
# SPEC-LOCAL-PLUS-PUBLISH: Local+ Published Notes and Privacy Tiers
|
||||
|
||||
**Status:** Draft
|
||||
**Date:** 2026-02-14
|
||||
**Owner:** Basic Memory
|
||||
|
||||
## Summary
|
||||
|
||||
Add a paid Local+ feature that lets users publish selected notes to shareable URLs while keeping the
|
||||
main knowledge base local-first. Use this as a product wedge for users who do not want full cloud
|
||||
hosting but do want collaboration and distribution features.
|
||||
|
||||
This spec also captures a practical position on "zero knowledge" for Local+.
|
||||
|
||||
## Context
|
||||
|
||||
Basic Memory already has strong local-first primitives and optional cloud routing/sync. A recurring
|
||||
request is:
|
||||
|
||||
- keep knowledge local by default,
|
||||
- pay for selective value-add,
|
||||
- share specific outputs externally.
|
||||
|
||||
Published Notes fits this model: explicit per-note opt-in, reversible, and easy to understand.
|
||||
|
||||
## Goals
|
||||
|
||||
1. Provide an Obsidian Publish-style sharing experience for selected notes.
|
||||
2. Keep local markdown files as source of truth.
|
||||
3. Make sharing compatible with current cloud/auth/billing primitives.
|
||||
4. Define clear Local+ packaging that does not degrade OSS local workflows.
|
||||
5. Document zero-knowledge constraints so product decisions are explicit.
|
||||
|
||||
## Non-Goals
|
||||
|
||||
1. Full hosted editing for all notes (Cloud Full remains separate).
|
||||
2. Public website builder/CMS features.
|
||||
3. Strict cryptographic zero-knowledge server processing for MCP/search in v1.
|
||||
|
||||
## Local+ Feature Catalog (Sellable)
|
||||
|
||||
Core Local+ candidates:
|
||||
|
||||
1. Published Notes (share URL, revoke, expiry, password).
|
||||
2. Snapshot Time Machine (point-in-time restore for local projects).
|
||||
3. Recovery Drill Reports (automated restore verification).
|
||||
4. Device/API Key Governance (per-device keys, revocation, audit trail).
|
||||
5. BYO Storage Orchestration (managed setup for user-owned object storage).
|
||||
6. Semantic Boost Add-on (higher quality retrieval options while files remain source-of-truth).
|
||||
|
||||
Team-oriented add-ons:
|
||||
|
||||
1. Team-owned shared links and domain branding.
|
||||
2. Role-based publish permissions.
|
||||
3. Shared workspace policies for what can be published.
|
||||
|
||||
## Proposed MVP: Published Notes
|
||||
|
||||
### User Experience
|
||||
|
||||
Per note actions:
|
||||
|
||||
1. Publish.
|
||||
2. Unpublish.
|
||||
3. Copy URL.
|
||||
4. Regenerate URL.
|
||||
5. Set visibility and controls.
|
||||
|
||||
Controls:
|
||||
|
||||
1. Visibility: `unlisted` (default) or `public`.
|
||||
2. Optional password gate.
|
||||
3. Optional expiration datetime.
|
||||
4. Optional "disable indexing" flag for public mode.
|
||||
|
||||
Behavior:
|
||||
|
||||
1. Source note remains local markdown.
|
||||
2. Publish is explicit opt-in per note.
|
||||
3. Unpublish removes public access immediately.
|
||||
4. Republish creates a new URL token unless user chooses to keep current URL.
|
||||
|
||||
### URL Model
|
||||
|
||||
1. Unlisted share URL: high-entropy token path.
|
||||
2. Public URL: slug path (optional, later phase).
|
||||
3. Team plans can support custom domain mapping in later phase.
|
||||
|
||||
### Content Model
|
||||
|
||||
v1 published page includes:
|
||||
|
||||
1. Rendered markdown body.
|
||||
2. Optional metadata (title, updated_at).
|
||||
|
||||
v1 excludes:
|
||||
|
||||
1. Full graph traversal expansion.
|
||||
2. Related note auto-discovery on public pages.
|
||||
|
||||
### Sync Model
|
||||
|
||||
1. Local file remains canonical.
|
||||
2. Publish stores a rendered snapshot plus metadata in cloud.
|
||||
3. Update path:
|
||||
- manual "update published version", or
|
||||
- optional auto-update on note change (plan-gated).
|
||||
|
||||
## Architecture (v1)
|
||||
|
||||
### High-Level Flow
|
||||
|
||||
1. Client selects a note to publish.
|
||||
2. Client sends publish request with note identifier and policy.
|
||||
3. Service resolves note content (local sync artifact or explicit upload payload).
|
||||
4. Service stores published artifact and returns share URL.
|
||||
|
||||
### Data Model
|
||||
|
||||
`published_notes`
|
||||
|
||||
1. `id` (uuid)
|
||||
2. `tenant_id` or `workspace_id`
|
||||
3. `project_id`
|
||||
4. `entity_permalink` (or stable external_id)
|
||||
5. `share_token` (hashed in DB)
|
||||
6. `visibility` (`unlisted`|`public`)
|
||||
7. `password_hash` (nullable)
|
||||
8. `expires_at` (nullable)
|
||||
9. `is_active`
|
||||
10. `published_content` (rendered snapshot or reference)
|
||||
11. `published_at`
|
||||
12. `updated_at`
|
||||
|
||||
### API Shape (Draft)
|
||||
|
||||
1. `POST /api/published-notes`
|
||||
2. `GET /api/published-notes`
|
||||
3. `GET /api/published-notes/{id}`
|
||||
4. `PATCH /api/published-notes/{id}`
|
||||
5. `DELETE /api/published-notes/{id}` (unpublish)
|
||||
6. `POST /api/published-notes/{id}/regenerate-url`
|
||||
7. `GET /p/{token}` (public resolver)
|
||||
|
||||
### CLI Shape (Draft)
|
||||
|
||||
1. `bm cloud publish <identifier>`
|
||||
2. `bm cloud publish list`
|
||||
3. `bm cloud publish update <id>`
|
||||
4. `bm cloud publish unpublish <id>`
|
||||
5. `bm cloud publish rotate-url <id>`
|
||||
|
||||
### Security
|
||||
|
||||
1. Default to unlisted URLs.
|
||||
2. Store only hashed share tokens.
|
||||
3. Passwords hashed server-side.
|
||||
4. Enforce expiration at request time.
|
||||
5. Log publish/unpublish/rotate events for auditability.
|
||||
|
||||
## Packaging and Pricing Direction
|
||||
|
||||
Suggested split:
|
||||
|
||||
1. OSS Local: no publish URLs.
|
||||
2. Local+ Solo: publish URLs + snapshots + recovery.
|
||||
3. Local+ Team: solo features + team governance and branding.
|
||||
4. Cloud Full: hosted app + full cloud workflows.
|
||||
|
||||
Key message:
|
||||
"Keep everything local. Publish only what you choose."
|
||||
|
||||
## Rollout Plan
|
||||
|
||||
1. Phase 1: Unlisted publish URLs + unpublish + regenerate URL.
|
||||
2. Phase 2: Password/expiry controls.
|
||||
3. Phase 3: Auto-update on note change and basic analytics.
|
||||
4. Phase 4: Team branding/domains/policies.
|
||||
|
||||
## Zero-Knowledge Position
|
||||
|
||||
### Strict Zero-Knowledge Definition
|
||||
|
||||
Strict zero-knowledge means the server cannot decrypt note content at all.
|
||||
|
||||
### Why This Conflicts with MCP and Search
|
||||
|
||||
If server cannot decrypt:
|
||||
|
||||
1. MCP tool execution against cloud content cannot read/write semantic content.
|
||||
2. Full-text search cannot index plaintext content.
|
||||
3. Semantic/vector search cannot generate or query embeddings on plaintext.
|
||||
4. Server-side relation resolution and context building become severely limited.
|
||||
|
||||
This matches earlier findings: strict zero-knowledge materially handicaps MCP-driven behavior and
|
||||
search quality.
|
||||
|
||||
### Viable Alternatives (Not Strict Zero-Knowledge)
|
||||
|
||||
1. Encryption at rest/in transit with server-side decrypt in trusted runtime.
|
||||
- Preserves MCP/search quality.
|
||||
- Not zero-knowledge cryptographically.
|
||||
|
||||
2. Client-side retrieval mode.
|
||||
- Keep MCP/search local; cloud is sync/share/backup relay.
|
||||
- Best for privacy-first users.
|
||||
- Requires local agent availability for advanced retrieval.
|
||||
|
||||
3. Limited encrypted indexing.
|
||||
- Blind indexes for exact keywords only.
|
||||
- No high-quality semantic search.
|
||||
- Usually poor UX for natural-language memory recall.
|
||||
|
||||
### Recommendation
|
||||
|
||||
For Local+:
|
||||
|
||||
1. Do not promise strict zero-knowledge for cloud MCP/search paths.
|
||||
2. Offer a privacy-first local mode where advanced retrieval stays local.
|
||||
3. Clearly label tradeoffs:
|
||||
- "Local private mode" (best privacy, best local retrieval).
|
||||
- "Cloud-assisted mode" (best cross-device/MCP consistency, trusted-runtime decrypt).
|
||||
|
||||
This keeps messaging honest and avoids repeating the known incompatibility.
|
||||
|
||||
@@ -1,368 +0,0 @@
|
||||
# SPEC-SCHEMA-IMPL: Schema System Implementation Plan
|
||||
|
||||
**Status:** Draft
|
||||
**Created:** 2025-02-06
|
||||
**Branch:** `feature/schema-system`
|
||||
**Depends on:** [SPEC-SCHEMA](SPEC-SCHEMA.md)
|
||||
|
||||
## Overview
|
||||
|
||||
Implementation plan for the Basic Memory Schema System. The system is entirely programmatic —
|
||||
no LLM agent runtime or API key required. The LLM already in the user's session (Claude Code,
|
||||
Claude Desktop, etc.) provides the intelligence layer by reading schema notes via existing
|
||||
MCP tools.
|
||||
|
||||
## Architecture
|
||||
|
||||
```
|
||||
┌─────────────────────────────────────────────────┐
|
||||
│ Entry Points │
|
||||
│ CLI (bm schema ...) │ MCP (schema_validate) │
|
||||
└──────────┬────────────┴──────────┬──────────────┘
|
||||
│ │
|
||||
▼ ▼
|
||||
┌─────────────────────────────────────────────────┐
|
||||
│ Schema Service Layer │
|
||||
│ resolve_schema · validate · infer · diff │
|
||||
└──────────┬────────────────────────┬──────────────┘
|
||||
│ │
|
||||
▼ ▼
|
||||
┌──────────────────────┐ ┌────────────────────────┐
|
||||
│ Picoschema Parser │ │ Note/Entity Access │
|
||||
│ YAML → SchemaModel │ │ (existing repository) │
|
||||
└──────────────────────┘ └────────────────────────┘
|
||||
```
|
||||
|
||||
No new database tables. Schemas are notes with `type: schema` — they're already indexed.
|
||||
Validation reads observations and relations from existing data.
|
||||
|
||||
## Components
|
||||
|
||||
### 1. Picoschema Parser
|
||||
|
||||
**Location:** `src/basic_memory/schema/parser.py`
|
||||
|
||||
Parses Picoschema YAML into an internal representation.
|
||||
|
||||
```python
|
||||
@dataclass
|
||||
class SchemaField:
|
||||
name: str
|
||||
type: str # string, integer, number, boolean, any, or EntityName
|
||||
required: bool # True unless field name ends with ?
|
||||
is_array: bool # True if (array) notation
|
||||
is_enum: bool # True if (enum) notation
|
||||
enum_values: list[str] # Populated for enums
|
||||
description: str | None # Text after comma
|
||||
is_entity_ref: bool # True if type is capitalized (entity reference)
|
||||
children: list[SchemaField] # For (object) types
|
||||
|
||||
|
||||
@dataclass
|
||||
class SchemaDefinition:
|
||||
entity: str # The entity type this schema describes
|
||||
version: int # Schema version
|
||||
fields: list[SchemaField] # Parsed fields
|
||||
validation_mode: str # "warn" | "strict" | "off"
|
||||
frontmatter_fields: list[SchemaField] # From settings.frontmatter (default: [])
|
||||
|
||||
|
||||
def parse_picoschema(yaml_dict: dict) -> list[SchemaField]:
|
||||
"""Parse a Picoschema YAML dict into a list of SchemaField objects."""
|
||||
|
||||
|
||||
def parse_schema_note(frontmatter: dict) -> SchemaDefinition:
|
||||
"""Parse a full schema note's frontmatter into a SchemaDefinition."""
|
||||
```
|
||||
|
||||
**Input/Output:**
|
||||
```yaml
|
||||
# Input (YAML dict from frontmatter)
|
||||
schema:
|
||||
name: string, full name
|
||||
role?: string, job title
|
||||
works_at?: Organization, employer
|
||||
expertise?(array): string, areas of knowledge
|
||||
```
|
||||
|
||||
```python
|
||||
# Output
|
||||
[
|
||||
SchemaField(name="name", type="string", required=True, description="full name", ...),
|
||||
SchemaField(name="role", type="string", required=False, description="job title", ...),
|
||||
SchemaField(name="works_at", type="Organization", required=False, is_entity_ref=True, ...),
|
||||
SchemaField(name="expertise", type="string", required=False, is_array=True, ...),
|
||||
]
|
||||
```
|
||||
|
||||
### 2. Schema Resolver
|
||||
|
||||
**Location:** `src/basic_memory/schema/resolver.py`
|
||||
|
||||
Finds the applicable schema for a note using the resolution order.
|
||||
|
||||
```python
|
||||
async def resolve_schema(
|
||||
note_frontmatter: dict,
|
||||
search_fn: Callable, # injected search capability
|
||||
) -> SchemaDefinition | None:
|
||||
"""Resolve schema for a note.
|
||||
|
||||
Resolution order:
|
||||
1. Inline schema (frontmatter['schema'] is a dict)
|
||||
2. Explicit reference (frontmatter['schema'] is a string)
|
||||
3. Implicit by type (frontmatter['type'] → schema note with matching entity)
|
||||
4. No schema (returns None)
|
||||
"""
|
||||
```
|
||||
|
||||
### 3. Schema Validator
|
||||
|
||||
**Location:** `src/basic_memory/schema/validator.py`
|
||||
|
||||
Validates a note's observations and relations against a resolved schema.
|
||||
|
||||
```python
|
||||
@dataclass
|
||||
class FieldResult:
|
||||
field: SchemaField
|
||||
status: str # "present" | "missing" | "type_mismatch"
|
||||
values: list[str] # Matched observation values or relation targets
|
||||
message: str | None # Human-readable detail
|
||||
|
||||
|
||||
@dataclass
|
||||
class ValidationResult:
|
||||
note_identifier: str
|
||||
schema_entity: str
|
||||
passed: bool # True if no errors (warnings are OK)
|
||||
field_results: list[FieldResult]
|
||||
unmatched_observations: dict[str, int] # category → count
|
||||
unmatched_relations: list[str] # relation types not in schema
|
||||
warnings: list[str]
|
||||
errors: list[str]
|
||||
|
||||
|
||||
async def validate_note(
|
||||
note: Note,
|
||||
schema: SchemaDefinition,
|
||||
frontmatter: dict | None = None,
|
||||
) -> ValidationResult:
|
||||
"""Validate a note against a schema definition.
|
||||
|
||||
Mapping rules:
|
||||
- field: string → observation [field] exists
|
||||
- field?(array): type → multiple [field] observations
|
||||
- field?: EntityType → relation 'field [[...]]' exists
|
||||
- field?(enum): [v] → observation [field] value ∈ enum values
|
||||
- settings.frontmatter field → frontmatter key presence/value
|
||||
"""
|
||||
```
|
||||
|
||||
### 4. Schema Inference Engine
|
||||
|
||||
**Location:** `src/basic_memory/schema/inference.py`
|
||||
|
||||
Analyzes notes of a given type and suggests a schema based on usage frequency.
|
||||
|
||||
```python
|
||||
@dataclass
|
||||
class FieldFrequency:
|
||||
name: str
|
||||
source: str # "observation" | "relation"
|
||||
count: int # notes containing this field
|
||||
total: int # total notes analyzed
|
||||
percentage: float
|
||||
sample_values: list[str] # representative values
|
||||
is_array: bool # True if typically appears multiple times per note
|
||||
target_type: str | None # For relations, the most common target entity type
|
||||
|
||||
|
||||
@dataclass
|
||||
class InferenceResult:
|
||||
entity_type: str
|
||||
notes_analyzed: int
|
||||
field_frequencies: list[FieldFrequency]
|
||||
suggested_schema: dict # Ready-to-use Picoschema YAML dict
|
||||
suggested_required: list[str]
|
||||
suggested_optional: list[str]
|
||||
excluded: list[str] # Below threshold
|
||||
|
||||
|
||||
async def infer_schema(
|
||||
entity_type: str,
|
||||
notes: list[Note],
|
||||
required_threshold: float = 0.95, # 95%+ = required
|
||||
optional_threshold: float = 0.25, # 25%+ = optional
|
||||
) -> InferenceResult:
|
||||
"""Analyze notes and suggest a Picoschema definition."""
|
||||
```
|
||||
|
||||
### 5. Schema Diff
|
||||
|
||||
**Location:** `src/basic_memory/schema/diff.py`
|
||||
|
||||
Compares current note usage against an existing schema definition.
|
||||
|
||||
```python
|
||||
@dataclass
|
||||
class SchemaDrift:
|
||||
new_fields: list[FieldFrequency] # Fields not in schema but common in notes
|
||||
dropped_fields: list[FieldFrequency] # Fields in schema but rare in notes
|
||||
cardinality_changes: list[str] # one → many or many → one
|
||||
type_mismatches: list[str] # observation values don't match declared type
|
||||
|
||||
|
||||
async def diff_schema(
|
||||
schema: SchemaDefinition,
|
||||
notes: list[Note],
|
||||
) -> SchemaDrift:
|
||||
"""Compare a schema against actual note usage to detect drift."""
|
||||
```
|
||||
|
||||
## Entry Points
|
||||
|
||||
### CLI Commands
|
||||
|
||||
**Location:** `src/basic_memory/cli/schema.py`
|
||||
|
||||
```python
|
||||
import typer
|
||||
|
||||
schema_app = typer.Typer(name="schema", help="Schema management commands")
|
||||
|
||||
@schema_app.command()
|
||||
async def validate(
|
||||
target: str = typer.Argument(None, help="Note path or entity type"),
|
||||
strict: bool = typer.Option(False, help="Override to strict mode"),
|
||||
):
|
||||
"""Validate notes against their schemas."""
|
||||
|
||||
@schema_app.command()
|
||||
async def infer(
|
||||
entity_type: str = typer.Argument(..., help="Entity type to analyze"),
|
||||
threshold: float = typer.Option(0.25, help="Minimum frequency for optional fields"),
|
||||
save: bool = typer.Option(False, help="Save to schema/ directory"),
|
||||
):
|
||||
"""Infer schema from existing notes of a type."""
|
||||
|
||||
@schema_app.command()
|
||||
async def diff(
|
||||
entity_type: str = typer.Argument(..., help="Entity type to diff"),
|
||||
):
|
||||
"""Show drift between schema and actual usage."""
|
||||
```
|
||||
|
||||
Registered as subcommand: `bm schema validate`, `bm schema infer`, `bm schema diff`.
|
||||
|
||||
### MCP Tools
|
||||
|
||||
**Location:** `src/basic_memory/mcp/tools/schema.py`
|
||||
|
||||
```python
|
||||
@mcp_tool
|
||||
async def schema_validate(
|
||||
entity_type: str | None = None,
|
||||
identifier: str | None = None,
|
||||
project: str | None = None,
|
||||
) -> str:
|
||||
"""Validate notes against their resolved schema."""
|
||||
|
||||
@mcp_tool
|
||||
async def schema_infer(
|
||||
entity_type: str,
|
||||
threshold: float = 0.25,
|
||||
project: str | None = None,
|
||||
) -> str:
|
||||
"""Analyze existing notes and suggest a schema definition."""
|
||||
```
|
||||
|
||||
### API Endpoints
|
||||
|
||||
**Location:** `src/basic_memory/api/schema_router.py`
|
||||
|
||||
```python
|
||||
router = APIRouter(prefix="/schema", tags=["schema"])
|
||||
|
||||
@router.post("/validate")
|
||||
async def validate_schema(...) -> ValidationReport: ...
|
||||
|
||||
@router.post("/infer")
|
||||
async def infer_schema(...) -> InferenceResult: ...
|
||||
|
||||
@router.get("/diff/{entity_type}")
|
||||
async def diff_schema(...) -> SchemaDrift: ...
|
||||
```
|
||||
|
||||
MCP tools call these endpoints via the typed client pattern (consistent with existing
|
||||
architecture).
|
||||
|
||||
## Implementation Phases
|
||||
|
||||
### Phase 1: Parser + Resolver
|
||||
|
||||
Build the foundation — can parse Picoschema and find schemas for notes.
|
||||
|
||||
**Deliverables:**
|
||||
- `schema/parser.py` — Picoschema YAML → `SchemaDefinition`
|
||||
- `schema/resolver.py` — Resolution order (inline → explicit ref → implicit by type → none)
|
||||
- Unit tests for all Picoschema syntax variations
|
||||
- Unit tests for resolution order
|
||||
|
||||
**No external dependencies.** Pure Python parsing of YAML dicts. Can develop and test
|
||||
in isolation.
|
||||
|
||||
### Phase 2: Validator
|
||||
|
||||
Connect schemas to notes and produce validation results.
|
||||
|
||||
**Deliverables:**
|
||||
- `schema/validator.py` — Validate note observations/relations against schema fields
|
||||
- API endpoint: `POST /schema/validate`
|
||||
- MCP tool: `schema_validate`
|
||||
- CLI command: `bm schema validate`
|
||||
- Integration tests with real notes and schemas
|
||||
|
||||
**Depends on:** Phase 1 (parser + resolver)
|
||||
|
||||
### Phase 3: Inference
|
||||
|
||||
Analyze existing notes to suggest schemas.
|
||||
|
||||
**Deliverables:**
|
||||
- `schema/inference.py` — Frequency analysis across notes of a type
|
||||
- API endpoint: `POST /schema/infer`
|
||||
- MCP tool: `schema_infer`
|
||||
- CLI command: `bm schema infer`
|
||||
- Option to save inferred schema as a note via `write_note`
|
||||
|
||||
**Depends on:** Phase 1 (parser for output format)
|
||||
|
||||
### Phase 4: Diff
|
||||
|
||||
Compare schemas against current usage.
|
||||
|
||||
**Deliverables:**
|
||||
- `schema/diff.py` — Drift detection between schema and actual notes
|
||||
- API endpoint: `GET /schema/diff/{entity_type}`
|
||||
- CLI command: `bm schema diff`
|
||||
|
||||
**Depends on:** Phase 1 (parser), Phase 3 (inference, for frequency analysis)
|
||||
|
||||
## Testing Strategy
|
||||
|
||||
- **Unit tests** (`tests/schema/`): Parser edge cases, resolution logic, validation mapping,
|
||||
inference thresholds
|
||||
- **Integration tests** (`test-int/schema/`): End-to-end with real markdown files, schema notes
|
||||
on disk, CLI invocation
|
||||
- Coverage target: 100% (consistent with project standard)
|
||||
|
||||
## What This Does NOT Include
|
||||
|
||||
- No new database tables or migrations
|
||||
- No new markdown syntax (schemas validate existing observations/relations)
|
||||
- No LLM agent runtime or API key management
|
||||
- No hook integration (deferred)
|
||||
- No schema composition/inheritance (deferred)
|
||||
- No OWL/RDF export (deferred)
|
||||
- No built-in templates (deferred)
|
||||
@@ -1,492 +0,0 @@
|
||||
# SPEC-SCHEMA: Basic Memory Schema System
|
||||
|
||||
**Status:** Draft
|
||||
**Created:** 2025-02-06
|
||||
**Branch:** `feature/schema-system`
|
||||
|
||||
## Summary
|
||||
|
||||
A schema system for Basic Memory that uses [Picoschema](https://genkit.dev/docs/dotprompt/)
|
||||
syntax in YAML frontmatter. Schemas validate notes against their existing observation/relation
|
||||
structure — no new data model, no migration, just a declarative lens over what's already there.
|
||||
|
||||
## Core Principles
|
||||
|
||||
1. **Schemas are just notes** — A schema is a note with `type: schema`, lives anywhere
|
||||
2. **Use prior art** — Picoschema syntax in YAML frontmatter, no custom notation
|
||||
3. **Validation maps to existing format** — Observations and relations, not a parallel data model
|
||||
4. **Validation is soft** — Warnings by default, not blocking errors
|
||||
5. **Inference over prescription** — Schemas describe reality, emerge from usage
|
||||
6. **No built-in agent** — Programmatic core; the LLM already in the session provides intelligence
|
||||
|
||||
## Picoschema Syntax
|
||||
|
||||
Picoschema is a compact schema notation from Google's Dotprompt that fits naturally in YAML
|
||||
frontmatter.
|
||||
|
||||
### Supported Types
|
||||
|
||||
| Type | Description |
|
||||
|------|-------------|
|
||||
| `string` | Text value |
|
||||
| `integer` | Whole number |
|
||||
| `number` | Decimal number |
|
||||
| `boolean` | True/false |
|
||||
| `any` | Any scalar type |
|
||||
| `EntityName` | Reference to another entity (capitalized = entity reference) |
|
||||
|
||||
### Syntax Rules
|
||||
|
||||
```yaml
|
||||
schema:
|
||||
name: string, full name # required field with description
|
||||
email?: string, contact email # ? = optional
|
||||
role?: string, job title
|
||||
works_at?: Organization, employer # capitalized type = entity reference
|
||||
tags?(array): string, categories # array of type
|
||||
status?(enum): [active, inactive] # enum with allowed values
|
||||
metadata?(object): # nested object
|
||||
updated_at?: string
|
||||
source?: string
|
||||
```
|
||||
|
||||
- `field: type` — required field
|
||||
- `field?: type` — optional field
|
||||
- `field(array): type` — array of values
|
||||
- `field?(enum): [values]` — enumeration
|
||||
- `field?(object):` — nested object with sub-fields
|
||||
- `, description` — description after comma
|
||||
- `EntityName` as type (capitalized) — reference to another entity
|
||||
|
||||
## Schema-to-Note Mapping
|
||||
|
||||
Schemas validate against the existing Basic Memory note format. No new syntax for note
|
||||
authors to learn.
|
||||
|
||||
### Mapping Rules
|
||||
|
||||
| Schema Declaration | Grounded In | Example Match |
|
||||
|--------------------|-------------|---------------|
|
||||
| `field: string` | Observation `[field] value` | `- [name] Paul Graham` |
|
||||
| `field?(array): string` | Multiple `[field]` observations | `- [expertise] Lisp` (×N) |
|
||||
| `field?: EntityType` | Relation `field [[Target]]` | `- works_at [[Y Combinator]]` |
|
||||
| `field?(array): EntityType` | Multiple `field` relations | `- authored [[Book]]` (×N) |
|
||||
| `tags` | Frontmatter `tags` array | `tags: [startups, essays]` |
|
||||
| `field?(enum): [values]` | Observation `[field] value` where value ∈ set | `- [status] active` |
|
||||
| `settings.frontmatter` field | Frontmatter key presence/value | `tags: [python, ai]` |
|
||||
|
||||
### Key Insight
|
||||
|
||||
Schemas don't introduce a new way to store data. They describe the patterns already present
|
||||
in observations and relations. A note doesn't have to change how it's written — the schema
|
||||
just says "a good Person note has a `[name]` observation and a `works_at` relation."
|
||||
|
||||
## Schema Definition
|
||||
|
||||
### As a Dedicated Schema Note
|
||||
|
||||
```yaml
|
||||
# schema/Person.md
|
||||
---
|
||||
title: Person
|
||||
type: schema
|
||||
entity: Person
|
||||
version: 1
|
||||
schema:
|
||||
name: string, full name
|
||||
email?: string, contact email
|
||||
role?: string, job title
|
||||
works_at?: Organization, employer
|
||||
expertise?(array): string, areas of knowledge
|
||||
settings:
|
||||
validation: warn # warn | strict | off
|
||||
frontmatter:
|
||||
tags?(array): string, note categories
|
||||
status?(enum): [draft, review, published]
|
||||
---
|
||||
|
||||
# Person
|
||||
|
||||
A human individual in the knowledge graph.
|
||||
|
||||
Any documentation about this entity type goes here as prose.
|
||||
```
|
||||
|
||||
Schema notes are regular Basic Memory notes. They show up in search, can have their own
|
||||
observations and relations, and can be organized in any folder (though `schema/` is
|
||||
the suggested convention).
|
||||
|
||||
### Inline Schema in a Note
|
||||
|
||||
Notes can carry their own schema directly:
|
||||
|
||||
```yaml
|
||||
# meetings/2024-01-15-standup.md
|
||||
---
|
||||
title: Team Standup 2024-01-15
|
||||
type: meeting
|
||||
schema:
|
||||
attendees(array): string, who was there
|
||||
decisions(array): string, what was decided
|
||||
action_items(array): string, follow-ups
|
||||
blockers?(array): string, anything stuck
|
||||
---
|
||||
|
||||
# Team Standup 2024-01-15
|
||||
|
||||
## Observations
|
||||
- [attendees] Paul
|
||||
- [attendees] Sarah
|
||||
- [decisions] Ship v2 by Friday
|
||||
- [action_items] Paul to review PR #42
|
||||
- [blockers] Waiting on API credentials
|
||||
```
|
||||
|
||||
Good for one-off structured notes or prototyping a schema before extracting it.
|
||||
|
||||
### Explicit Schema Reference
|
||||
|
||||
A note can reference a schema by entity name or permalink:
|
||||
|
||||
```yaml
|
||||
# projects/basic-memory.md
|
||||
---
|
||||
title: Basic Memory
|
||||
schema: SoftwareProject # by entity name
|
||||
---
|
||||
|
||||
# research/llm-memory-patterns.md
|
||||
---
|
||||
title: LLM Memory Patterns
|
||||
schema: schema/research-project # by permalink
|
||||
---
|
||||
```
|
||||
|
||||
Use cases:
|
||||
- Note's `type` differs from the schema it should validate against
|
||||
- Multiple schema variants exist for the same domain
|
||||
- Applying structure to existing notes without changing their type
|
||||
|
||||
## Schema Resolution
|
||||
|
||||
When validating a note, schemas resolve in priority order:
|
||||
|
||||
```
|
||||
1. Inline schema → schema: { ... } (dict in frontmatter)
|
||||
2. Explicit ref → schema: Person (string in frontmatter)
|
||||
3. Implicit by type → type: Person (lookup schema note with entity: Person)
|
||||
4. No schema → no validation (perfectly fine)
|
||||
```
|
||||
|
||||
```python
|
||||
async def resolve_schema(note: Note) -> Schema | None:
|
||||
schema_value = note.frontmatter.get('schema')
|
||||
|
||||
# 1. Inline schema (dict)
|
||||
if isinstance(schema_value, dict):
|
||||
return parse_picoschema(schema_value)
|
||||
|
||||
# 2. Explicit reference (string)
|
||||
if isinstance(schema_value, str):
|
||||
schema_note = await find_schema_note(schema_value)
|
||||
if schema_note:
|
||||
return parse_picoschema(schema_note.frontmatter['schema'])
|
||||
|
||||
# 3. Implicit by type
|
||||
note_type = note.frontmatter.get('type')
|
||||
if note_type:
|
||||
results = await search_notes(f"type:schema entity:{note_type}")
|
||||
if results:
|
||||
return parse_picoschema(results[0].frontmatter['schema'])
|
||||
|
||||
# 4. No schema
|
||||
return None
|
||||
```
|
||||
|
||||
## Validation
|
||||
|
||||
### Modes
|
||||
|
||||
Configured in the schema's `settings.validation`:
|
||||
|
||||
| Mode | Behavior |
|
||||
|------|----------|
|
||||
| `off` | No validation |
|
||||
| `warn` | Warnings in output, doesn't block (default) |
|
||||
| `strict` | Errors that block sync, for CI/CD enforcement |
|
||||
|
||||
### Validation Output
|
||||
|
||||
For a note missing required fields:
|
||||
|
||||
```
|
||||
$ bm schema validate people/ada-lovelace.md
|
||||
|
||||
⚠ Person schema validation:
|
||||
- Missing required field: name (expected [name] observation)
|
||||
- Missing optional field: role
|
||||
- Missing optional field: works_at (no relation found)
|
||||
|
||||
ℹ Unmatched observations: [fact] ×2, [born] ×1
|
||||
ℹ Unmatched relations: collaborated_with
|
||||
```
|
||||
|
||||
"Unmatched" items are informational — observations and relations the schema doesn't cover.
|
||||
They're valid. Schemas are a subset, not a straitjacket.
|
||||
|
||||
### Frontmatter Validation
|
||||
|
||||
Schema notes can declare validation rules for frontmatter keys under `settings.frontmatter`
|
||||
using the same Picoschema syntax as the `schema` block:
|
||||
|
||||
```yaml
|
||||
settings:
|
||||
validation: warn
|
||||
frontmatter:
|
||||
tags?(array): string
|
||||
status?(enum): [draft, review, published]
|
||||
```
|
||||
|
||||
- Frontmatter rules use the same Picoschema key syntax (`?` for optional, `(enum)`, `(array)`)
|
||||
- Only available on schema notes (inline schemas skip frontmatter validation)
|
||||
- Checks key presence (required vs optional) and enum value membership
|
||||
- Unmatched frontmatter keys not in the schema are silently ignored
|
||||
- Missing required frontmatter keys produce a warning (or error in strict mode)
|
||||
|
||||
Example output for a missing required frontmatter key:
|
||||
|
||||
```
|
||||
⚠ Person schema validation:
|
||||
- Missing required frontmatter key: status
|
||||
```
|
||||
|
||||
### Batch Validation
|
||||
|
||||
```
|
||||
$ bm schema validate Person
|
||||
|
||||
Validating 30 notes against Person schema...
|
||||
|
||||
✓ people/paul-graham.md — all fields present
|
||||
✓ people/rich-hickey.md — all fields present
|
||||
⚠ people/ada-lovelace.md — missing: name
|
||||
⚠ people/alan-kay.md — missing: name, role
|
||||
✓ people/linus-torvalds.md — all fields present
|
||||
...
|
||||
|
||||
Summary: 22/30 valid, 8 warnings, 0 errors
|
||||
```
|
||||
|
||||
## Emerging Schemas
|
||||
|
||||
### The Problem with Traditional Schemas
|
||||
|
||||
Most schema systems require: define schema → create conforming content → fight the schema
|
||||
when reality doesn't match. This is backwards. Knowledge grows organically.
|
||||
|
||||
### The Basic Memory Approach
|
||||
|
||||
```
|
||||
Write notes freely → Patterns emerge → Crystallize into schema → Validate future notes
|
||||
```
|
||||
|
||||
### Schema Inference
|
||||
|
||||
Generate schemas from existing notes by analyzing observation and relation frequency:
|
||||
|
||||
```
|
||||
$ bm schema infer Person
|
||||
|
||||
Analyzing 30 notes with type: Person...
|
||||
|
||||
Observations found:
|
||||
[name] 30/30 100% → name: string
|
||||
[role] 27/30 90% → role?: string
|
||||
[fact] 25/30 83% (generic — no single field)
|
||||
[expertise] 18/30 60% → expertise?(array): string
|
||||
[email] 8/30 27% → email?: string
|
||||
[born] 6/30 20% (below threshold)
|
||||
|
||||
Relations found:
|
||||
works_at 22/30 73% → works_at?: Organization
|
||||
authored 11/30 37% → authored?(array): string
|
||||
|
||||
Suggested schema:
|
||||
name: string, full name
|
||||
role?: string, job title
|
||||
expertise?(array): string, areas of knowledge
|
||||
email?: string, contact email
|
||||
works_at?: Organization, employer
|
||||
|
||||
Save to schema/Person.md? [y/n]
|
||||
```
|
||||
|
||||
Frequency thresholds:
|
||||
- 100% present → required field
|
||||
- 25%+ present → optional field
|
||||
- Below 25% → excluded from suggestion (but noted)
|
||||
|
||||
### Schema Drift Detection
|
||||
|
||||
Track how usage patterns shift over time:
|
||||
|
||||
```
|
||||
$ bm schema diff Person
|
||||
|
||||
Schema drift detected:
|
||||
|
||||
+ expertise: now in 81% of notes (was 12%)
|
||||
- department: dropped to 3% of notes
|
||||
~ works_at: cardinality changed (one → many)
|
||||
|
||||
Update schema? [y/n/review]
|
||||
```
|
||||
|
||||
## LLM Integration (AI Guidance)
|
||||
|
||||
No agent runtime or API key required. The LLM already in the session uses schemas as
|
||||
context for note creation.
|
||||
|
||||
### Flow
|
||||
|
||||
1. User asks LLM to "write a note about Rich Hickey"
|
||||
2. LLM determines `type: Person` is appropriate
|
||||
3. LLM calls `search_notes("type:schema entity:Person")` → finds schema
|
||||
4. LLM reads schema fields: required `name`, optional `role`, `works_at`, `expertise`
|
||||
5. LLM calls `write_note` with observations and relations that satisfy the schema
|
||||
|
||||
The schema acts as a creation template. The LLM knows what a "complete" note looks like
|
||||
without any custom agent infrastructure.
|
||||
|
||||
### MCP Tools
|
||||
|
||||
```python
|
||||
@mcp_tool
|
||||
async def schema_validate(
|
||||
entity_type: str | None = None,
|
||||
identifier: str | None = None,
|
||||
project: str | None = None,
|
||||
) -> ValidationReport:
|
||||
"""Validate notes against their resolved schema.
|
||||
|
||||
Validates a specific note (by identifier) or all notes of a given type.
|
||||
Returns warnings/errors based on the schema's validation mode.
|
||||
"""
|
||||
|
||||
@mcp_tool
|
||||
async def schema_infer(
|
||||
entity_type: str,
|
||||
threshold: float = 0.25,
|
||||
project: str | None = None,
|
||||
) -> SuggestedSchema:
|
||||
"""Analyze existing notes and suggest a schema definition.
|
||||
|
||||
Examines observation categories and relation types across all notes
|
||||
of the given type. Returns frequency analysis and suggested Picoschema.
|
||||
"""
|
||||
```
|
||||
|
||||
## CLI Commands
|
||||
|
||||
```bash
|
||||
# Validate a specific note
|
||||
bm schema validate people/ada-lovelace.md
|
||||
|
||||
# Validate all notes of a type
|
||||
bm schema validate Person
|
||||
|
||||
# Validate everything with a schema
|
||||
bm schema validate
|
||||
|
||||
# Infer schema from existing notes
|
||||
bm schema infer Person
|
||||
|
||||
# Show schema drift from current definition
|
||||
bm schema diff Person
|
||||
|
||||
# List all schema notes
|
||||
bm search "type:schema"
|
||||
```
|
||||
|
||||
## Examples
|
||||
|
||||
### Complete Person Workflow
|
||||
|
||||
**Schema:**
|
||||
```yaml
|
||||
# schema/Person.md
|
||||
---
|
||||
title: Person
|
||||
type: schema
|
||||
entity: Person
|
||||
version: 1
|
||||
schema:
|
||||
name: string, full name
|
||||
role?: string, job title or position
|
||||
works_at?: Organization, employer
|
||||
expertise?(array): string, areas of knowledge
|
||||
email?: string, contact email
|
||||
settings:
|
||||
validation: warn
|
||||
---
|
||||
|
||||
# Person
|
||||
|
||||
A human individual in the knowledge graph.
|
||||
```
|
||||
|
||||
**Valid note:**
|
||||
```yaml
|
||||
# people/paul-graham.md
|
||||
---
|
||||
title: Paul Graham
|
||||
type: Person
|
||||
tags: [startups, essays, lisp]
|
||||
---
|
||||
|
||||
# Paul Graham
|
||||
|
||||
## Observations
|
||||
- [name] Paul Graham
|
||||
- [role] Essayist and investor
|
||||
- [expertise] Startups
|
||||
- [expertise] Lisp
|
||||
- [expertise] Essay writing
|
||||
- [fact] Created Viaweb, the first web app
|
||||
|
||||
## Relations
|
||||
- works_at [[Y Combinator]]
|
||||
- authored [[Hackers and Painters]]
|
||||
```
|
||||
|
||||
**Note with warnings:**
|
||||
```yaml
|
||||
# people/ada-lovelace.md
|
||||
---
|
||||
title: Ada Lovelace
|
||||
type: Person
|
||||
---
|
||||
|
||||
# Ada Lovelace
|
||||
|
||||
## Observations
|
||||
- [fact] Wrote the first computer program
|
||||
- [born] 1815
|
||||
|
||||
## Relations
|
||||
- collaborated_with [[Charles Babbage]]
|
||||
```
|
||||
|
||||
Validation: warns about missing required `[name]` observation. Everything else is optional
|
||||
or unmatched (which is fine).
|
||||
|
||||
## Future Considerations (Deferred)
|
||||
|
||||
These are interesting but out of scope for the initial implementation:
|
||||
|
||||
- **Multiple schema inheritance** — `schema: [Person, Author]`
|
||||
- **Hook integration** — Pre-write validation via the hooks system
|
||||
- **OWL/RDF export** — `bm schema export --format owl`
|
||||
- **SPARQL queries** — Schema-aware graph queries
|
||||
- **Built-in templates** — `bm schema use gtd`, `bm schema use zettelkasten`
|
||||
- **Schema versioning/migration** — Tracking breaking changes across versions
|
||||
@@ -1,28 +0,0 @@
|
||||
## Coverage policy (practical 100%)
|
||||
|
||||
Basic Memory’s test suite intentionally mixes:
|
||||
- unit tests (fast, deterministic)
|
||||
- integration tests (real filesystem + real DB via `test-int/`)
|
||||
|
||||
To keep the default CI signal **stable and meaningful**, the default `pytest` coverage report targets **core library logic** and **excludes** a small set of modules that are either:
|
||||
- highly environment-dependent (OS/DB tuning)
|
||||
- inherently interactive (CLI)
|
||||
- background-task orchestration (watchers/sync runners)
|
||||
|
||||
### What's excluded (and why)
|
||||
|
||||
Coverage excludes are configured in `pyproject.toml` under `[tool.coverage.report].omit`.
|
||||
|
||||
Current exclusions include:
|
||||
- `src/basic_memory/cli/**`: interactive wrappers; behavior is validated via higher-level tests and smoke tests.
|
||||
- `src/basic_memory/db.py`: platform/backend tuning paths (SQLite/Postgres/Windows), covered by integration tests and targeted runs.
|
||||
- `src/basic_memory/services/initialization.py`: startup orchestration/background tasks; covered indirectly by app/MCP entrypoints.
|
||||
- `src/basic_memory/sync/sync_service.py`: heavy filesystem↔DB integration; validated in integration suite (not enforced in unit coverage).
|
||||
|
||||
### Recommended additional runs
|
||||
|
||||
If you want extra confidence locally/CI:
|
||||
- **Postgres backend**: run tests with `BASIC_MEMORY_TEST_POSTGRES=1`.
|
||||
- **Strict backend-complete coverage**: run coverage on SQLite + Postgres and combine the results (recommended).
|
||||
|
||||
|
||||
@@ -1,7 +0,0 @@
|
||||
{
|
||||
"$schema": "https://glama.ai/mcp/schemas/server.json",
|
||||
"maintainers": [
|
||||
"phernandez",
|
||||
"groksrc"
|
||||
]
|
||||
}
|
||||
@@ -1,9 +0,0 @@
|
||||
[run]
|
||||
source = .
|
||||
omit =
|
||||
tests/*
|
||||
tests/**/*
|
||||
|
||||
[report]
|
||||
fail_under = 85
|
||||
show_missing = True
|
||||
@@ -1,48 +0,0 @@
|
||||
name: integration
|
||||
|
||||
# Heavier than test.yml — installs the real `basic-memory` CLI via uv, runs
|
||||
# every bm_* tool against a live `bm mcp` subprocess. Catches BM-API drift
|
||||
# (e.g., a bm release renaming a tool argument) before our users see it.
|
||||
concurrency:
|
||||
group: hbm-integration-${{ github.workflow }}-${{ github.ref }}
|
||||
cancel-in-progress: true
|
||||
|
||||
on:
|
||||
push:
|
||||
branches: [main]
|
||||
workflow_dispatch:
|
||||
|
||||
jobs:
|
||||
integration:
|
||||
name: Integration tests (real bm + mcp)
|
||||
runs-on: ubuntu-latest
|
||||
timeout-minutes: 15
|
||||
|
||||
steps:
|
||||
- uses: actions/checkout@v4
|
||||
|
||||
- name: Install uv
|
||||
# No cache (no uv.lock to key off — see test.yml comment).
|
||||
uses: astral-sh/setup-uv@v3
|
||||
|
||||
- name: Set up Python 3.12
|
||||
# basic-memory itself requires 3.12+; the bm install needs that.
|
||||
# Hermes-runtime compatibility (3.11) is covered by test.yml.
|
||||
run: uv python install 3.12
|
||||
|
||||
- name: Install basic-memory CLI via uv
|
||||
run: |
|
||||
uv tool install basic-memory
|
||||
# uv puts entry-point shims under ~/.local/bin
|
||||
echo "$HOME/.local/bin" >> "$GITHUB_PATH"
|
||||
|
||||
- name: Verify bm is on PATH
|
||||
run: |
|
||||
which bm
|
||||
bm --version
|
||||
|
||||
- name: Run integration tests
|
||||
env:
|
||||
BM_INTEGRATION: "1"
|
||||
run: |
|
||||
uv run --with pytest --with mcp --python 3.12 pytest tests/test_integration.py -v
|
||||
@@ -1,36 +0,0 @@
|
||||
name: pr-title
|
||||
|
||||
on:
|
||||
pull_request:
|
||||
types: [opened, edited, synchronize]
|
||||
|
||||
jobs:
|
||||
semantic-pr-title:
|
||||
runs-on: ubuntu-latest
|
||||
steps:
|
||||
- uses: amannn/action-semantic-pull-request@v5
|
||||
env:
|
||||
GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
|
||||
with:
|
||||
# Conventional-commit types we accept in PR titles + commit subjects.
|
||||
types: |
|
||||
feat
|
||||
fix
|
||||
chore
|
||||
docs
|
||||
style
|
||||
refactor
|
||||
perf
|
||||
test
|
||||
build
|
||||
ci
|
||||
# Single-file plugin — no real submodule structure. We don't require
|
||||
# a scope, but if a contributor uses one we accept these:
|
||||
scopes: |
|
||||
core
|
||||
tests
|
||||
ci
|
||||
docs
|
||||
deps
|
||||
requireScope: false
|
||||
requireScopeForBreakingChange: true
|
||||
-194
@@ -1,194 +0,0 @@
|
||||
name: release
|
||||
|
||||
# Manual trigger: Actions → release → Run workflow. Always runs against main
|
||||
# (the workflow validates GITHUB_REF). Steps:
|
||||
# 1. Compute the new version from the `version` input (patch/minor/major
|
||||
# or explicit semver). The *current* version is read from the latest
|
||||
# git tag (`v*.*.*`) — NOT from __init__.py. This is robust to PRs that
|
||||
# pre-bump __init__.py: the bump always runs from the last released
|
||||
# version, not from whatever the working files happen to say.
|
||||
# 2. Update __version__ in __init__.py and version in plugin.yaml to match
|
||||
# the new tag — bringing the files in sync if a PR pre-bumped them.
|
||||
# 3. Commit as `chore(release): vX.Y.Z`, tag, push to main + push the tag.
|
||||
# 4. Publish a GitHub Release. Body is the matching `## [X.Y.Z]` block from
|
||||
# CHANGELOG.md when present; otherwise auto-generated release notes.
|
||||
#
|
||||
# Recommended flow: land a PR that adds a `## [X.Y.Z]` section to CHANGELOG.md
|
||||
# first, then run this workflow with the matching version so the release notes
|
||||
# are the hand-written changelog instead of commit-message-derived notes.
|
||||
on:
|
||||
workflow_dispatch:
|
||||
inputs:
|
||||
version:
|
||||
description: "Version bump (`patch`, `minor`, `major`) or explicit semver (`0.3.0`)"
|
||||
required: true
|
||||
default: "patch"
|
||||
|
||||
permissions:
|
||||
contents: write
|
||||
|
||||
concurrency:
|
||||
group: release-${{ github.ref }}
|
||||
cancel-in-progress: false
|
||||
|
||||
jobs:
|
||||
release:
|
||||
name: Tag and Publish GitHub Release
|
||||
runs-on: ubuntu-latest
|
||||
steps:
|
||||
- name: Validate trigger is main
|
||||
run: |
|
||||
if [ "$GITHUB_REF" != "refs/heads/main" ]; then
|
||||
echo "::error::release must run against main. Got $GITHUB_REF"
|
||||
exit 1
|
||||
fi
|
||||
|
||||
- uses: actions/checkout@v4
|
||||
with:
|
||||
fetch-depth: 0
|
||||
token: ${{ secrets.GITHUB_TOKEN }}
|
||||
|
||||
- name: Compute new version
|
||||
id: bump
|
||||
run: |
|
||||
set -euo pipefail
|
||||
VERSION_INPUT="${{ github.event.inputs.version }}"
|
||||
|
||||
# Current = latest released tag (sort by version, descending; pick
|
||||
# the first v*.*.* tag). NOT __init__.py — a PR may have pre-bumped
|
||||
# the version files, and we don't want to double-bump on top of
|
||||
# that. The bump is computed from the last *released* version.
|
||||
LAST_TAG=$(git tag --list 'v*.*.*' --sort=-v:refname | head -1 || true)
|
||||
if [ -z "$LAST_TAG" ]; then
|
||||
# First release in the repo. Seed with 0.0.0 so a `patch` bump
|
||||
# yields v0.0.1, `minor` yields v0.1.0, `major` yields v1.0.0.
|
||||
# Explicit semver inputs bypass the seed entirely.
|
||||
CURRENT="0.0.0"
|
||||
echo "No prior v*.*.* tag found; seeding current=0.0.0"
|
||||
else
|
||||
CURRENT="${LAST_TAG#v}"
|
||||
echo "Latest released tag: $LAST_TAG (current=$CURRENT)"
|
||||
fi
|
||||
|
||||
if [[ "$VERSION_INPUT" =~ ^(patch|minor|major)$ ]]; then
|
||||
IFS=. read -r MAJ MIN PAT <<< "$CURRENT"
|
||||
case "$VERSION_INPUT" in
|
||||
major) MAJ=$((MAJ + 1)); MIN=0; PAT=0 ;;
|
||||
minor) MIN=$((MIN + 1)); PAT=0 ;;
|
||||
patch) PAT=$((PAT + 1)) ;;
|
||||
esac
|
||||
NEW="${MAJ}.${MIN}.${PAT}"
|
||||
elif [[ "$VERSION_INPUT" =~ ^[0-9]+\.[0-9]+\.[0-9]+$ ]]; then
|
||||
NEW="$VERSION_INPUT"
|
||||
else
|
||||
echo "::error::Invalid version input: '$VERSION_INPUT'"
|
||||
echo "::error::Use patch|minor|major or explicit X.Y.Z"
|
||||
exit 1
|
||||
fi
|
||||
|
||||
if [ "$NEW" = "$CURRENT" ]; then
|
||||
echo "::error::New version equals last released ($NEW). Pick a different version."
|
||||
exit 1
|
||||
fi
|
||||
|
||||
# Sanity check: refuse if __init__.py is already ahead of the
|
||||
# version we're about to ship. Catches the "PR bumped to 0.5.0 but
|
||||
# workflow was asked for a patch that would land 0.1.8" foot-gun
|
||||
# before it overwrites the files.
|
||||
FILE_VERSION=$(grep -E '^__version__ = ' __init__.py \
|
||||
| sed -E 's/^__version__ = "([^"]+)".*/\1/')
|
||||
if [ -n "$FILE_VERSION" ] && [ "$FILE_VERSION" != "$CURRENT" ] && [ "$FILE_VERSION" != "$NEW" ]; then
|
||||
echo "::error::__init__.py reports version $FILE_VERSION, but the release would ship $NEW (last tag: $CURRENT)."
|
||||
echo "::error::Reconcile by picking a version input that matches __init__.py, or roll __init__.py back to $CURRENT."
|
||||
exit 1
|
||||
fi
|
||||
|
||||
echo "New version: $NEW"
|
||||
echo "current=$CURRENT" >> "$GITHUB_OUTPUT"
|
||||
echo "version=$NEW" >> "$GITHUB_OUTPUT"
|
||||
echo "tag=v$NEW" >> "$GITHUB_OUTPUT"
|
||||
|
||||
- name: Refuse if tag already exists
|
||||
run: |
|
||||
set -euo pipefail
|
||||
TAG="${{ steps.bump.outputs.tag }}"
|
||||
if git rev-parse --verify "refs/tags/$TAG" >/dev/null 2>&1; then
|
||||
echo "::error::Tag $TAG already exists locally. Pick a different version."
|
||||
exit 1
|
||||
fi
|
||||
if git ls-remote --tags origin "$TAG" | grep -q "refs/tags/$TAG$"; then
|
||||
echo "::error::Tag $TAG already exists on origin. Pick a different version."
|
||||
exit 1
|
||||
fi
|
||||
|
||||
- name: Update version files
|
||||
run: |
|
||||
set -euo pipefail
|
||||
NEW="${{ steps.bump.outputs.version }}"
|
||||
sed -i -E "s/^__version__ = \"[^\"]+\"/__version__ = \"${NEW}\"/" __init__.py
|
||||
sed -i -E "s/^version: .*/version: ${NEW}/" plugin.yaml
|
||||
|
||||
# Verify both files changed and that the new version is present.
|
||||
grep -q "^__version__ = \"${NEW}\"" __init__.py
|
||||
grep -q "^version: ${NEW}$" plugin.yaml
|
||||
|
||||
echo "--- diff ---"
|
||||
git --no-pager diff -- __init__.py plugin.yaml
|
||||
|
||||
- name: Configure Git identity
|
||||
run: |
|
||||
git config user.name "github-actions[bot]"
|
||||
git config user.email "41898282+github-actions[bot]@users.noreply.github.com"
|
||||
|
||||
- name: Commit and tag
|
||||
run: |
|
||||
set -euo pipefail
|
||||
TAG="${{ steps.bump.outputs.tag }}"
|
||||
git add __init__.py plugin.yaml
|
||||
if git diff --cached --quiet; then
|
||||
# PR already bumped the files to the target version. Tag the
|
||||
# existing HEAD rather than creating an empty release commit.
|
||||
echo "Version files already at ${TAG}; tagging current HEAD."
|
||||
else
|
||||
git commit -m "chore(release): ${TAG}"
|
||||
fi
|
||||
git tag -a "${TAG}" -m "${TAG}"
|
||||
|
||||
- name: Push commit and tag
|
||||
run: |
|
||||
set -euo pipefail
|
||||
# HEAD push is a no-op when nothing was committed in this run.
|
||||
git push origin HEAD:main
|
||||
git push origin "${{ steps.bump.outputs.tag }}"
|
||||
|
||||
- name: Extract CHANGELOG section for this version
|
||||
id: changelog
|
||||
run: |
|
||||
set -euo pipefail
|
||||
VERSION="${{ steps.bump.outputs.version }}"
|
||||
# Pull lines between `## [VERSION]` and the next `## [` heading.
|
||||
SECTION=$(awk -v ver="$VERSION" '
|
||||
$0 ~ "^## \\[" ver "\\]" { found=1; next }
|
||||
found && /^## \[/ { exit }
|
||||
found { print }
|
||||
' CHANGELOG.md)
|
||||
if [ -z "$SECTION" ]; then
|
||||
echo "::warning::No CHANGELOG.md section found for v${VERSION} — falling back to auto-generated release notes."
|
||||
echo "has_section=false" >> "$GITHUB_OUTPUT"
|
||||
else
|
||||
echo "has_section=true" >> "$GITHUB_OUTPUT"
|
||||
{
|
||||
echo 'body<<EOF_CHANGELOG'
|
||||
echo "$SECTION"
|
||||
echo 'EOF_CHANGELOG'
|
||||
} >> "$GITHUB_OUTPUT"
|
||||
fi
|
||||
|
||||
- name: Create GitHub Release
|
||||
uses: softprops/action-gh-release@v2
|
||||
with:
|
||||
tag_name: ${{ steps.bump.outputs.tag }}
|
||||
name: ${{ steps.bump.outputs.tag }}
|
||||
body: ${{ steps.changelog.outputs.body }}
|
||||
generate_release_notes: ${{ steps.changelog.outputs.has_section == 'false' }}
|
||||
token: ${{ secrets.GITHUB_TOKEN }}
|
||||
-39
@@ -1,39 +0,0 @@
|
||||
name: tests
|
||||
|
||||
# Cancel an in-progress run when a new commit lands on the same branch — the
|
||||
# latest result is the one we care about.
|
||||
concurrency:
|
||||
group: hbm-tests-${{ github.workflow }}-${{ github.ref }}
|
||||
cancel-in-progress: true
|
||||
|
||||
on:
|
||||
# Branch pushes already cover PRs (the PR branch tip is what's being tested),
|
||||
# so we don't run the matrix twice for the same commit on push + pull_request.
|
||||
push:
|
||||
|
||||
jobs:
|
||||
unit:
|
||||
name: Unit tests (Python ${{ matrix.python-version }})
|
||||
runs-on: ubuntu-latest
|
||||
timeout-minutes: 10
|
||||
strategy:
|
||||
fail-fast: false
|
||||
matrix:
|
||||
# 3.11 is the runtime Hermes itself ships on today; 3.12-3.14 cover
|
||||
# forward-compat for whenever Hermes upgrades.
|
||||
python-version: ["3.11", "3.12", "3.13", "3.14"]
|
||||
|
||||
steps:
|
||||
- uses: actions/checkout@v4
|
||||
|
||||
- name: Install uv
|
||||
# No `enable-cache: true` — the action's default cache key globs for
|
||||
# `uv.lock`, which we don't ship (we use `uv run --with` instead of
|
||||
# `uv sync`). Without a lock file the cache step errors out.
|
||||
uses: astral-sh/setup-uv@v3
|
||||
|
||||
- name: Set up Python ${{ matrix.python-version }}
|
||||
run: uv python install ${{ matrix.python-version }}
|
||||
|
||||
- name: Run unit tests
|
||||
run: uv run --with pytest --python ${{ matrix.python-version }} pytest -q
|
||||
@@ -1,6 +0,0 @@
|
||||
__pycache__/
|
||||
*.pyc
|
||||
.venv/
|
||||
.DS_Store
|
||||
.pytest_cache/
|
||||
*.egg-info/
|
||||
@@ -1,133 +0,0 @@
|
||||
# Changelog
|
||||
|
||||
All notable changes to this project will be documented in this file.
|
||||
|
||||
The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.1.0/), and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
|
||||
|
||||
## [0.3.2] — 2026-05-23
|
||||
|
||||
### Fixed
|
||||
- **Let Basic Memory v0.21.3 self-route workspace-qualified identifiers and URLs.** Hermes no longer injects its configured default project into `bm_read`, `bm_edit`, `bm_delete`, `bm_move`, or `bm_context` calls when the identifier/URL is already workspace-qualified, such as `personal/main/...`, `memory://personal/main/...`, or an organization workspace slug with a 32-character hash suffix. This preserves Basic Memory Cloud's workspace-aware routing while keeping the existing default-project behavior for short/local identifiers.
|
||||
|
||||
## [0.3.1] — 2026-05-16
|
||||
|
||||
### Changed
|
||||
- **Documented the Hermes Agent v0.14.0-compatible `/bm-*` slash-command monkeypatch.** `MONKEYPATCH.md` now distinguishes the plugin's runtime version from the Hermes Agent-side compatibility patch: plugin `v0.3.0` remains the correct runtime release for Hermes Agent `v0.13.x`, while Hermes Agent `v0.14.0` still needs the updated two-part core patch so gateway startup command discovery loads the active exclusive memory provider and the memory-provider collector delegates `register_command` / `register_skill`.
|
||||
- **Clarified install guidance for users and agents.** The README known-issue section now points Hermes `v0.13.x` and `v0.14.0` users at the compatibility matrix in `MONKEYPATCH.md`, so agents do not mistake a plugin update for the required Hermes Agent core patch.
|
||||
|
||||
### Notes
|
||||
- This is a documentation/compatibility-instructions release only. It does not change the plugin runtime code or Basic Memory data behavior. The plugin remains backward-compatible with Hermes Agent `v0.13.x`; the new documentation explains how to patch Hermes Agent `v0.14.0` until the upstream Hermes fix ships.
|
||||
|
||||
## [0.3.0] — 2026-05-12
|
||||
|
||||
### Added
|
||||
- **Per-call project routing on every `bm_*` tool.** All eight tools now accept optional `project` (name) and `project_id` (UUID from `bm_projects`) parameters. The agent can write or read against a project other than the Hermes-configured one — useful when the user asks to write into a different cloud project (e.g. a personal `main` project) without reconfiguring the plugin. `project_id` takes precedence over `project`; both fall back to the configured default when omitted. Workspace routing is handled transparently by BM via `project_id` — no separate workspace parameter is needed.
|
||||
- **`bm_projects` and `bm_workspaces` agent tools.** Promotes the discovery logic previously available only as `/bm-project` and `/bm-workspace` slash commands to agent-facing tools. `bm_projects` returns JSON with `name` and `external_id` (UUID) per project so the agent can hand the UUID to `bm_write` / `bm_read` / etc. via `project_id` — the unambiguous form across cloud workspaces. `bm_workspaces` lists BM Cloud workspaces (name, type, role, default flag). Together with per-call routing, these unblock the workflow Drew's friction note flagged: agent picks the right project + workspace before writing, instead of silently operating against the active Hermes memory project.
|
||||
- **SKILL.md cross-project workflow** documenting the discovery → route → write → verify recipe end-to-end. Adds a "Permalinks" section covering the three canonical shapes (short, project-qualified, workspace-qualified) and the round-trip property where `bm_write`'s returned permalink self-routes for follow-up reads. A "Cross-project routing" section explains `project` (including workspace-qualified syntax like `"personal/main"`) vs `project_id` and when to use each. Also backfills `bm_recent` documentation (the tool shipped in 0.2.0 but the skill hadn't been updated).
|
||||
- **SKILL.md "Further reading" section** linking to the official docs at [docs.basicmemory.com](https://docs.basicmemory.com), with raw-markdown URLs (`/raw/<path>.md`) the agent can `WebFetch` on demand for deeper material — knowledge format, observations & relations, memory URL wildcards, semantic search, cloud routing, BM's full MCP tool surface, and the `llms.txt` sitemap.
|
||||
|
||||
### Notes
|
||||
- Addresses the routing, discovery, and documentation gaps in the real-world note "Hermes Basic Memory Cloud Task Experience." A proposed `bm_import` tool was evaluated and dropped — `read_file` + `bm_write` already composes the same operation with no new capability, at the cost of one more tool in the surface.
|
||||
- The slash commands `/bm-project` and `/bm-workspace` still exist and behave identically — they continue to call `list_memory_projects` / `list_workspaces` directly via the actor. No behavior change for human use.
|
||||
|
||||
## [0.2.0] — 2026-05-11
|
||||
|
||||
### Added
|
||||
- **Plugin-owned `/bm-*` slash commands** for CLI/gateway sessions. Eight commands give humans direct memory-graph access without going through the agent: `/bm-search`, `/bm-read`, `/bm-context`, `/bm-recent`, `/bm-status`, `/bm-remember`, `/bm-project`, `/bm-workspace`. Closes #2.
|
||||
- **`bm_recent` tool** wrapping BM's `recent_activity`. Surfaces notes updated within a timeframe (`7d` default, accepts natural language like `"2 weeks"` or `"yesterday"`). Agent-facing and reused by `/bm-recent`.
|
||||
- **`remember_folder` config key** (default `"bm-remember"`). Separate from `capture_folder` so manual captures via `/bm-remember` don't intermix with auto-generated session transcripts. Notes are tagged `manual-capture` for further disambiguation.
|
||||
|
||||
### Fixed
|
||||
- **`ctx.register_skill(...)` was silently no-opping since 0.1.5** in real Hermes installs. Hermes loads memory-provider plugins through a stripped-down `_ProviderCollector` context (`plugins/memory/__init__.py`) that captures only `register_memory_provider`; `register_skill` and `register_command` are not delegated. The plugin now writes directly to `PluginManager._plugin_commands` and `_plugin_skills`, matching the entry shape and name normalization `PluginContext.register_command` / `register_skill` produce. This makes both the new slash commands and the bundled SKILL.md work in current Hermes installs. The clean fix lives upstream — a small patch to teach `_ProviderCollector` to delegate — and once that lands, the reach-in becomes a redundant double-write of identical entries. Forward-compat `ctx.register_command` / `ctx.register_skill` calls remain in place for the future code path.
|
||||
|
||||
### Notes
|
||||
- `/bm-remember` derives the title from the first non-empty line of the input, trimmed to 80 chars; falls back to `Note YYYY-MM-DD HHMM UTC`.
|
||||
- `/bm-workspace` short-circuits in local mode with a one-line explanation. Workspaces are a BM Cloud concept.
|
||||
- Mid-session project/workspace switching is intentionally not supported in 0.2.0 — auto-capture would land in unexpected places. Tracked as a follow-up.
|
||||
|
||||
## [0.1.7] — 2026-05-10
|
||||
|
||||
### Changed
|
||||
- **Stronger nudge in `system_prompt_block()`** to steer agents toward the `bm_*` tools instead of shelling out to `bm` CLI. Pre-v0.1.7 the prompt listed the tools neutrally; given Claude/Hermes models' heavy training-data exposure to `bm tool ...` CLI patterns, neutral language wasn't enough — agents reached for the shell by reflex, paying 1-2s of cold-start per call instead of ~0.1s through our persistent MCP connection. New prompt is explicit (**"Use the `bm_*` tools below directly — do not shell out to the `bm` CLI"**) and gives a one-line latency rationale so the model has a reason to follow it.
|
||||
- `SKILL.md` mirrors the directive with a "Use `bm_*`, not the `bm` CLI" section + a tool-vs-CLI table.
|
||||
|
||||
### Added
|
||||
- Regression test `test_system_prompt_block_steers_away_from_cli` locks in the directive language so future prompt edits don't accidentally weaken it.
|
||||
|
||||
## [0.1.6] — 2026-05-10
|
||||
|
||||
### Fixed
|
||||
- **`bm_*` tools were never registered with Hermes's `MemoryManager._tool_to_provider`.** `get_tool_schemas()` was gated on `self._initialized`, but Hermes captures the schema list at *register* time — before `initialize()` runs. The gate caused every session to start with zero tools registered for our provider, so every LLM-issued `bm_search` (and friends) returned `"Unknown tool: bm_search"` from MemoryManager's dispatch. Symptoms were asymmetric: prefetch (recall injection) worked because it's invoked per-turn after init, but tool calls didn't. Schemas are static — they now return unconditionally, with `handle_tool_call()` doing the runtime "is the actor ready?" gate.
|
||||
- Regression test pins this so we don't reintroduce it: `test_get_tool_schemas_unconditional` asserts `get_tool_schemas()` returns all 7 schemas on a fresh, uninitialized provider.
|
||||
|
||||
## [0.1.5] — 2026-05-10
|
||||
|
||||
### Added
|
||||
- Bundled `SKILL.md` is now auto-registered via `ctx.register_skill("basic-memory", ...)` during plugin load. No more manual symlink to `~/.hermes/skills/`. The skill is opt-in (resolvable via `skill:view basic-memory:basic-memory`); always-on agent guidance still flows through `system_prompt_block()`.
|
||||
|
||||
### Changed
|
||||
- README rewritten for community install. Lead command is now `hermes plugins install basicmachines-co/hermes-basic-memory`. Clone-and-symlink instructions moved to the Development section.
|
||||
- Added GitHub Actions CI: unit tests on push and PR.
|
||||
- Added this CHANGELOG.
|
||||
|
||||
## [0.1.4] — 2026-05-10
|
||||
|
||||
### Added
|
||||
- `_uv_binary_path()` and `_install_bm_via_uv()`. When `bm` is missing from the host, the plugin runs `uv tool install basic-memory --quiet` once at first `initialize()`. The bm binary lands at `~/.local/bin/bm` — the same canonical path a manual `uv tool install basic-memory` produces, so subsequent manual installs are no-ops rather than creating a second install.
|
||||
- 8 new unit tests covering `is_available()` with bm/uv combinations, the install subprocess (success / non-zero exit / OSError / no-uv), and `initialize()` install-or-not branching.
|
||||
|
||||
### Changed
|
||||
- `is_available()` now returns `True` when **either** `bm` is on disk **or** `uv` is on disk (we can install the missing CLI ourselves).
|
||||
- README's prerequisites section: dropped manual basic-memory install requirement; added the one-time ~10s cold-start note.
|
||||
|
||||
## [0.1.3] — 2026-05-10
|
||||
|
||||
### Fixed
|
||||
- README's cloud-mode section described the wrong setup (`bm project add ... --cloud --local-path` + `bm cloud bisync`), which gives a local-mode project with file-level cloud sync rather than true cloud routing. Replaced with `bm project set-cloud <name> --workspace <name>`, which flips the project to `ProjectMode.CLOUD` so tool calls route over HTTPS to `<cloud_host>/proxy` directly. No local files involved.
|
||||
- Documented OAuth / API-key auth options, and the `--workspace` requirement when the user belongs to multiple BM Cloud workspaces.
|
||||
|
||||
## [0.1.2] — 2026-05-10
|
||||
|
||||
### Changed
|
||||
- `_default_project()`: `"hermes-memory"` (was `"hermes-{hostname}"`).
|
||||
- `_default_project_path()`: `~/hermes-memory/` (was `~/.basic-memory/hermes/`). The previous path violated the principle that `~/.basic-memory/` is reserved for BM's app state, not project storage.
|
||||
|
||||
### Added
|
||||
- `_bm_known_projects()` reads bm's `~/.basic-memory/config.json`. `BasicMemoryProvider._verify_project_registered()` uses it to refuse initialization when `mode: cloud` is set against a project that isn't registered with bm. Local mode still auto-creates as before.
|
||||
- 13 new unit tests for the introspection + bail-out paths.
|
||||
|
||||
## [0.1.1] — 2026-05-10
|
||||
|
||||
### Added
|
||||
- `tests/test_actor.py` — 15 tests covering `_BmMcpActor` lifecycle, call dispatch, timeout-with-cancellation, idempotent shutdown.
|
||||
- `tests/test_capture.py` — 25 tests for `sync_turn` (first-write + append paths), `on_session_end` summary shape, and gating.
|
||||
- `tests/test_prefetch.py` — 25 tests for `prefetch` / `queue_prefetch` / `_format_prefetch` including forward-compat with unknown response fields.
|
||||
- `tests/test_integration.py` — 12 gated tests exercising every tool against a real `bm` MCP server (`BM_INTEGRATION=1` + `bm` + `mcp`). Each session uses a throwaway BM project that's torn down on completion.
|
||||
|
||||
### Changed
|
||||
- `_BmMcpActor.call` now refuses calls after `shutdown()` (sets `_running=False`) and cancels the underlying coroutine on timeout instead of leaking it.
|
||||
- `_format_prefetch` defensively coerces non-string fields and skips non-dict entries.
|
||||
- Added module-level `__version__`, kept in sync with `plugin.yaml` (verified by a test).
|
||||
|
||||
## [0.1.0] — 2026-05-10
|
||||
|
||||
### Added
|
||||
- Initial release of the Hermes Memory Provider plugin for Basic Memory.
|
||||
- Seven `bm_*` agent tools: `bm_search`, `bm_read`, `bm_write`, `bm_edit`, `bm_context`, `bm_delete`, `bm_move`.
|
||||
- Per-turn capture (`sync_turn`) and end-of-session summary (`on_session_end`).
|
||||
- Local mode (default) with auto-created BM project; cloud mode with project-name-based routing.
|
||||
- Single-file plugin at `__init__.py`, AGPL-3.0-or-later.
|
||||
- 84-test pytest suite.
|
||||
|
||||
[0.3.2]: https://github.com/basicmachines-co/hermes-basic-memory/releases/tag/v0.3.2
|
||||
[0.3.1]: https://github.com/basicmachines-co/hermes-basic-memory/releases/tag/v0.3.1
|
||||
[0.3.0]: https://github.com/basicmachines-co/hermes-basic-memory/releases/tag/v0.3.0
|
||||
[0.2.0]: https://github.com/basicmachines-co/hermes-basic-memory/releases/tag/v0.2.0
|
||||
[0.1.7]: https://github.com/basicmachines-co/hermes-basic-memory/releases/tag/v0.1.7
|
||||
[0.1.6]: https://github.com/basicmachines-co/hermes-basic-memory/releases/tag/v0.1.6
|
||||
[0.1.5]: https://github.com/basicmachines-co/hermes-basic-memory/releases/tag/v0.1.5
|
||||
[0.1.4]: https://github.com/basicmachines-co/hermes-basic-memory/releases/tag/v0.1.4
|
||||
[0.1.3]: https://github.com/basicmachines-co/hermes-basic-memory/releases/tag/v0.1.3
|
||||
[0.1.2]: https://github.com/basicmachines-co/hermes-basic-memory/releases/tag/v0.1.2
|
||||
[0.1.1]: https://github.com/basicmachines-co/hermes-basic-memory/releases/tag/v0.1.1
|
||||
[0.1.0]: https://github.com/basicmachines-co/hermes-basic-memory/releases/tag/v0.1.0
|
||||
@@ -1,661 +0,0 @@
|
||||
GNU AFFERO GENERAL PUBLIC LICENSE
|
||||
Version 3, 19 November 2007
|
||||
|
||||
Copyright (C) 2007 Free Software Foundation, Inc. <https://fsf.org/>
|
||||
Everyone is permitted to copy and distribute verbatim copies
|
||||
of this license document, but changing it is not allowed.
|
||||
|
||||
Preamble
|
||||
|
||||
The GNU Affero General Public License is a free, copyleft license for
|
||||
software and other kinds of works, specifically designed to ensure
|
||||
cooperation with the community in the case of network server software.
|
||||
|
||||
The licenses for most software and other practical works are designed
|
||||
to take away your freedom to share and change the works. By contrast,
|
||||
our General Public Licenses are intended to guarantee your freedom to
|
||||
share and change all versions of a program--to make sure it remains free
|
||||
software for all its users.
|
||||
|
||||
When we speak of free software, we are referring to freedom, not
|
||||
price. Our General Public Licenses are designed to make sure that you
|
||||
have the freedom to distribute copies of free software (and charge for
|
||||
them if you wish), that you receive source code or can get it if you
|
||||
want it, that you can change the software or use pieces of it in new
|
||||
free programs, and that you know you can do these things.
|
||||
|
||||
Developers that use our General Public Licenses protect your rights
|
||||
with two steps: (1) assert copyright on the software, and (2) offer
|
||||
you this License which gives you legal permission to copy, distribute
|
||||
and/or modify the software.
|
||||
|
||||
A secondary benefit of defending all users' freedom is that
|
||||
improvements made in alternate versions of the program, if they
|
||||
receive widespread use, become available for other developers to
|
||||
incorporate. Many developers of free software are heartened and
|
||||
encouraged by the resulting cooperation. However, in the case of
|
||||
software used on network servers, this result may fail to come about.
|
||||
The GNU General Public License permits making a modified version and
|
||||
letting the public access it on a server without ever releasing its
|
||||
source code to the public.
|
||||
|
||||
The GNU Affero General Public License is designed specifically to
|
||||
ensure that, in such cases, the modified source code becomes available
|
||||
to the community. It requires the operator of a network server to
|
||||
provide the source code of the modified version running there to the
|
||||
users of that server. Therefore, public use of a modified version, on
|
||||
a publicly accessible server, gives the public access to the source
|
||||
code of the modified version.
|
||||
|
||||
An older license, called the Affero General Public License and
|
||||
published by Affero, was designed to accomplish similar goals. This is
|
||||
a different license, not a version of the Affero GPL, but Affero has
|
||||
released a new version of the Affero GPL which permits relicensing under
|
||||
this license.
|
||||
|
||||
The precise terms and conditions for copying, distribution and
|
||||
modification follow.
|
||||
|
||||
TERMS AND CONDITIONS
|
||||
|
||||
0. Definitions.
|
||||
|
||||
"This License" refers to version 3 of the GNU Affero General Public License.
|
||||
|
||||
"Copyright" also means copyright-like laws that apply to other kinds of
|
||||
works, such as semiconductor masks.
|
||||
|
||||
"The Program" refers to any copyrightable work licensed under this
|
||||
License. Each licensee is addressed as "you". "Licensees" and
|
||||
"recipients" may be individuals or organizations.
|
||||
|
||||
To "modify" a work means to copy from or adapt all or part of the work
|
||||
in a fashion requiring copyright permission, other than the making of an
|
||||
exact copy. The resulting work is called a "modified version" of the
|
||||
earlier work or a work "based on" the earlier work.
|
||||
|
||||
A "covered work" means either the unmodified Program or a work based
|
||||
on the Program.
|
||||
|
||||
To "propagate" a work means to do anything with it that, without
|
||||
permission, would make you directly or secondarily liable for
|
||||
infringement under applicable copyright law, except executing it on a
|
||||
computer or modifying a private copy. Propagation includes copying,
|
||||
distribution (with or without modification), making available to the
|
||||
public, and in some countries other activities as well.
|
||||
|
||||
To "convey" a work means any kind of propagation that enables other
|
||||
parties to make or receive copies. Mere interaction with a user through
|
||||
a computer network, with no transfer of a copy, is not conveying.
|
||||
|
||||
An interactive user interface displays "Appropriate Legal Notices"
|
||||
to the extent that it includes a convenient and prominently visible
|
||||
feature that (1) displays an appropriate copyright notice, and (2)
|
||||
tells the user that there is no warranty for the work (except to the
|
||||
extent that warranties are provided), that licensees may convey the
|
||||
work under this License, and how to view a copy of this License. If
|
||||
the interface presents a list of user commands or options, such as a
|
||||
menu, a prominent item in the list meets this criterion.
|
||||
|
||||
1. Source Code.
|
||||
|
||||
The "source code" for a work means the preferred form of the work
|
||||
for making modifications to it. "Object code" means any non-source
|
||||
form of a work.
|
||||
|
||||
A "Standard Interface" means an interface that either is an official
|
||||
standard defined by a recognized standards body, or, in the case of
|
||||
interfaces specified for a particular programming language, one that
|
||||
is widely used among developers working in that language.
|
||||
|
||||
The "System Libraries" of an executable work include anything, other
|
||||
than the work as a whole, that (a) is included in the normal form of
|
||||
packaging a Major Component, but which is not part of that Major
|
||||
Component, and (b) serves only to enable use of the work with that
|
||||
Major Component, or to implement a Standard Interface for which an
|
||||
implementation is available to the public in source code form. A
|
||||
"Major Component", in this context, means a major essential component
|
||||
(kernel, window system, and so on) of the specific operating system
|
||||
(if any) on which the executable work runs, or a compiler used to
|
||||
produce the work, or an object code interpreter used to run it.
|
||||
|
||||
The "Corresponding Source" for a work in object code form means all
|
||||
the source code needed to generate, install, and (for an executable
|
||||
work) run the object code and to modify the work, including scripts to
|
||||
control those activities. However, it does not include the work's
|
||||
System Libraries, or general-purpose tools or generally available free
|
||||
programs which are used unmodified in performing those activities but
|
||||
which are not part of the work. For example, Corresponding Source
|
||||
includes interface definition files associated with source files for
|
||||
the work, and the source code for shared libraries and dynamically
|
||||
linked subprograms that the work is specifically designed to require,
|
||||
such as by intimate data communication or control flow between those
|
||||
subprograms and other parts of the work.
|
||||
|
||||
The Corresponding Source need not include anything that users
|
||||
can regenerate automatically from other parts of the Corresponding
|
||||
Source.
|
||||
|
||||
The Corresponding Source for a work in source code form is that
|
||||
same work.
|
||||
|
||||
2. Basic Permissions.
|
||||
|
||||
All rights granted under this License are granted for the term of
|
||||
copyright on the Program, and are irrevocable provided the stated
|
||||
conditions are met. This License explicitly affirms your unlimited
|
||||
permission to run the unmodified Program. The output from running a
|
||||
covered work is covered by this License only if the output, given its
|
||||
content, constitutes a covered work. This License acknowledges your
|
||||
rights of fair use or other equivalent, as provided by copyright law.
|
||||
|
||||
You may make, run and propagate covered works that you do not
|
||||
convey, without conditions so long as your license otherwise remains
|
||||
in force. You may convey covered works to others for the sole purpose
|
||||
of having them make modifications exclusively for you, or provide you
|
||||
with facilities for running those works, provided that you comply with
|
||||
the terms of this License in conveying all material for which you do
|
||||
not control copyright. Those thus making or running the covered works
|
||||
for you must do so exclusively on your behalf, under your direction
|
||||
and control, on terms that prohibit them from making any copies of
|
||||
your copyrighted material outside their relationship with you.
|
||||
|
||||
Conveying under any other circumstances is permitted solely under
|
||||
the conditions stated below. Sublicensing is not allowed; section 10
|
||||
makes it unnecessary.
|
||||
|
||||
3. Protecting Users' Legal Rights From Anti-Circumvention Law.
|
||||
|
||||
No covered work shall be deemed part of an effective technological
|
||||
measure under any applicable law fulfilling obligations under article
|
||||
11 of the WIPO copyright treaty adopted on 20 December 1996, or
|
||||
similar laws prohibiting or restricting circumvention of such
|
||||
measures.
|
||||
|
||||
When you convey a covered work, you waive any legal power to forbid
|
||||
circumvention of technological measures to the extent such circumvention
|
||||
is effected by exercising rights under this License with respect to
|
||||
the covered work, and you disclaim any intention to limit operation or
|
||||
modification of the work as a means of enforcing, against the work's
|
||||
users, your or third parties' legal rights to forbid circumvention of
|
||||
technological measures.
|
||||
|
||||
4. Conveying Verbatim Copies.
|
||||
|
||||
You may convey verbatim copies of the Program's source code as you
|
||||
receive it, in any medium, provided that you conspicuously and
|
||||
appropriately publish on each copy an appropriate copyright notice;
|
||||
keep intact all notices stating that this License and any
|
||||
non-permissive terms added in accord with section 7 apply to the code;
|
||||
keep intact all notices of the absence of any warranty; and give all
|
||||
recipients a copy of this License along with the Program.
|
||||
|
||||
You may charge any price or no price for each copy that you convey,
|
||||
and you may offer support or warranty protection for a fee.
|
||||
|
||||
5. Conveying Modified Source Versions.
|
||||
|
||||
You may convey a work based on the Program, or the modifications to
|
||||
produce it from the Program, in the form of source code under the
|
||||
terms of section 4, provided that you also meet all of these conditions:
|
||||
|
||||
a) The work must carry prominent notices stating that you modified
|
||||
it, and giving a relevant date.
|
||||
|
||||
b) The work must carry prominent notices stating that it is
|
||||
released under this License and any conditions added under section
|
||||
7. This requirement modifies the requirement in section 4 to
|
||||
"keep intact all notices".
|
||||
|
||||
c) You must license the entire work, as a whole, under this
|
||||
License to anyone who comes into possession of a copy. This
|
||||
License will therefore apply, along with any applicable section 7
|
||||
additional terms, to the whole of the work, and all its parts,
|
||||
regardless of how they are packaged. This License gives no
|
||||
permission to license the work in any other way, but it does not
|
||||
invalidate such permission if you have separately received it.
|
||||
|
||||
d) If the work has interactive user interfaces, each must display
|
||||
Appropriate Legal Notices; however, if the Program has interactive
|
||||
interfaces that do not display Appropriate Legal Notices, your
|
||||
work need not make them do so.
|
||||
|
||||
A compilation of a covered work with other separate and independent
|
||||
works, which are not by their nature extensions of the covered work,
|
||||
and which are not combined with it such as to form a larger program,
|
||||
in or on a volume of a storage or distribution medium, is called an
|
||||
"aggregate" if the compilation and its resulting copyright are not
|
||||
used to limit the access or legal rights of the compilation's users
|
||||
beyond what the individual works permit. Inclusion of a covered work
|
||||
in an aggregate does not cause this License to apply to the other
|
||||
parts of the aggregate.
|
||||
|
||||
6. Conveying Non-Source Forms.
|
||||
|
||||
You may convey a covered work in object code form under the terms
|
||||
of sections 4 and 5, provided that you also convey the
|
||||
machine-readable Corresponding Source under the terms of this License,
|
||||
in one of these ways:
|
||||
|
||||
a) Convey the object code in, or embodied in, a physical product
|
||||
(including a physical distribution medium), accompanied by the
|
||||
Corresponding Source fixed on a durable physical medium
|
||||
customarily used for software interchange.
|
||||
|
||||
b) Convey the object code in, or embodied in, a physical product
|
||||
(including a physical distribution medium), accompanied by a
|
||||
written offer, valid for at least three years and valid for as
|
||||
long as you offer spare parts or customer support for that product
|
||||
model, to give anyone who possesses the object code either (1) a
|
||||
copy of the Corresponding Source for all the software in the
|
||||
product that is covered by this License, on a durable physical
|
||||
medium customarily used for software interchange, for a price no
|
||||
more than your reasonable cost of physically performing this
|
||||
conveying of source, or (2) access to copy the
|
||||
Corresponding Source from a network server at no charge.
|
||||
|
||||
c) Convey individual copies of the object code with a copy of the
|
||||
written offer to provide the Corresponding Source. This
|
||||
alternative is allowed only occasionally and noncommercially, and
|
||||
only if you received the object code with such an offer, in accord
|
||||
with subsection 6b.
|
||||
|
||||
d) Convey the object code by offering access from a designated
|
||||
place (gratis or for a charge), and offer equivalent access to the
|
||||
Corresponding Source in the same way through the same place at no
|
||||
further charge. You need not require recipients to copy the
|
||||
Corresponding Source along with the object code. If the place to
|
||||
copy the object code is a network server, the Corresponding Source
|
||||
may be on a different server (operated by you or a third party)
|
||||
that supports equivalent copying facilities, provided you maintain
|
||||
clear directions next to the object code saying where to find the
|
||||
Corresponding Source. Regardless of what server hosts the
|
||||
Corresponding Source, you remain obligated to ensure that it is
|
||||
available for as long as needed to satisfy these requirements.
|
||||
|
||||
e) Convey the object code using peer-to-peer transmission, provided
|
||||
you inform other peers where the object code and Corresponding
|
||||
Source of the work are being offered to the general public at no
|
||||
charge under subsection 6d.
|
||||
|
||||
A separable portion of the object code, whose source code is excluded
|
||||
from the Corresponding Source as a System Library, need not be
|
||||
included in conveying the object code work.
|
||||
|
||||
A "User Product" is either (1) a "consumer product", which means any
|
||||
tangible personal property which is normally used for personal, family,
|
||||
or household purposes, or (2) anything designed or sold for incorporation
|
||||
into a dwelling. In determining whether a product is a consumer product,
|
||||
doubtful cases shall be resolved in favor of coverage. For a particular
|
||||
product received by a particular user, "normally used" refers to a
|
||||
typical or common use of that class of product, regardless of the status
|
||||
of the particular user or of the way in which the particular user
|
||||
actually uses, or expects or is expected to use, the product. A product
|
||||
is a consumer product regardless of whether the product has substantial
|
||||
commercial, industrial or non-consumer uses, unless such uses represent
|
||||
the only significant mode of use of the product.
|
||||
|
||||
"Installation Information" for a User Product means any methods,
|
||||
procedures, authorization keys, or other information required to install
|
||||
and execute modified versions of a covered work in that User Product from
|
||||
a modified version of its Corresponding Source. The information must
|
||||
suffice to ensure that the continued functioning of the modified object
|
||||
code is in no case prevented or interfered with solely because
|
||||
modification has been made.
|
||||
|
||||
If you convey an object code work under this section in, or with, or
|
||||
specifically for use in, a User Product, and the conveying occurs as
|
||||
part of a transaction in which the right of possession and use of the
|
||||
User Product is transferred to the recipient in perpetuity or for a
|
||||
fixed term (regardless of how the transaction is characterized), the
|
||||
Corresponding Source conveyed under this section must be accompanied
|
||||
by the Installation Information. But this requirement does not apply
|
||||
if neither you nor any third party retains the ability to install
|
||||
modified object code on the User Product (for example, the work has
|
||||
been installed in ROM).
|
||||
|
||||
The requirement to provide Installation Information does not include a
|
||||
requirement to continue to provide support service, warranty, or updates
|
||||
for a work that has been modified or installed by the recipient, or for
|
||||
the User Product in which it has been modified or installed. Access to a
|
||||
network may be denied when the modification itself materially and
|
||||
adversely affects the operation of the network or violates the rules and
|
||||
protocols for communication across the network.
|
||||
|
||||
Corresponding Source conveyed, and Installation Information provided,
|
||||
in accord with this section must be in a format that is publicly
|
||||
documented (and with an implementation available to the public in
|
||||
source code form), and must require no special password or key for
|
||||
unpacking, reading or copying.
|
||||
|
||||
7. Additional Terms.
|
||||
|
||||
"Additional permissions" are terms that supplement the terms of this
|
||||
License by making exceptions from one or more of its conditions.
|
||||
Additional permissions that are applicable to the entire Program shall
|
||||
be treated as though they were included in this License, to the extent
|
||||
that they are valid under applicable law. If additional permissions
|
||||
apply only to part of the Program, that part may be used separately
|
||||
under those permissions, but the entire Program remains governed by
|
||||
this License without regard to the additional permissions.
|
||||
|
||||
When you convey a copy of a covered work, you may at your option
|
||||
remove any additional permissions from that copy, or from any part of
|
||||
it. (Additional permissions may be written to require their own
|
||||
removal in certain cases when you modify the work.) You may place
|
||||
additional permissions on material, added by you to a covered work,
|
||||
for which you have or can give appropriate copyright permission.
|
||||
|
||||
Notwithstanding any other provision of this License, for material you
|
||||
add to a covered work, you may (if authorized by the copyright holders of
|
||||
that material) supplement the terms of this License with terms:
|
||||
|
||||
a) Disclaiming warranty or limiting liability differently from the
|
||||
terms of sections 15 and 16 of this License; or
|
||||
|
||||
b) Requiring preservation of specified reasonable legal notices or
|
||||
author attributions in that material or in the Appropriate Legal
|
||||
Notices displayed by works containing it; or
|
||||
|
||||
c) Prohibiting misrepresentation of the origin of that material, or
|
||||
requiring that modified versions of such material be marked in
|
||||
reasonable ways as different from the original version; or
|
||||
|
||||
d) Limiting the use for publicity purposes of names of licensors or
|
||||
authors of the material; or
|
||||
|
||||
e) Declining to grant rights under trademark law for use of some
|
||||
trade names, trademarks, or service marks; or
|
||||
|
||||
f) Requiring indemnification of licensors and authors of that
|
||||
material by anyone who conveys the material (or modified versions of
|
||||
it) with contractual assumptions of liability to the recipient, for
|
||||
any liability that these contractual assumptions directly impose on
|
||||
those licensors and authors.
|
||||
|
||||
All other non-permissive additional terms are considered "further
|
||||
restrictions" within the meaning of section 10. If the Program as you
|
||||
received it, or any part of it, contains a notice stating that it is
|
||||
governed by this License along with a term that is a further
|
||||
restriction, you may remove that term. If a license document contains
|
||||
a further restriction but permits relicensing or conveying under this
|
||||
License, you may add to a covered work material governed by the terms
|
||||
of that license document, provided that the further restriction does
|
||||
not survive such relicensing or conveying.
|
||||
|
||||
If you add terms to a covered work in accord with this section, you
|
||||
must place, in the relevant source files, a statement of the
|
||||
additional terms that apply to those files, or a notice indicating
|
||||
where to find the applicable terms.
|
||||
|
||||
Additional terms, permissive or non-permissive, may be stated in the
|
||||
form of a separately written license, or stated as exceptions;
|
||||
the above requirements apply either way.
|
||||
|
||||
8. Termination.
|
||||
|
||||
You may not propagate or modify a covered work except as expressly
|
||||
provided under this License. Any attempt otherwise to propagate or
|
||||
modify it is void, and will automatically terminate your rights under
|
||||
this License (including any patent licenses granted under the third
|
||||
paragraph of section 11).
|
||||
|
||||
However, if you cease all violation of this License, then your
|
||||
license from a particular copyright holder is reinstated (a)
|
||||
provisionally, unless and until the copyright holder explicitly and
|
||||
finally terminates your license, and (b) permanently, if the copyright
|
||||
holder fails to notify you of the violation by some reasonable means
|
||||
prior to 60 days after the cessation.
|
||||
|
||||
Moreover, your license from a particular copyright holder is
|
||||
reinstated permanently if the copyright holder notifies you of the
|
||||
violation by some reasonable means, this is the first time you have
|
||||
received notice of violation of this License (for any work) from that
|
||||
copyright holder, and you cure the violation prior to 30 days after
|
||||
your receipt of the notice.
|
||||
|
||||
Termination of your rights under this section does not terminate the
|
||||
licenses of parties who have received copies or rights from you under
|
||||
this License. If your rights have been terminated and not permanently
|
||||
reinstated, you do not qualify to receive new licenses for the same
|
||||
material under section 10.
|
||||
|
||||
9. Acceptance Not Required for Having Copies.
|
||||
|
||||
You are not required to accept this License in order to receive or
|
||||
run a copy of the Program. Ancillary propagation of a covered work
|
||||
occurring solely as a consequence of using peer-to-peer transmission
|
||||
to receive a copy likewise does not require acceptance. However,
|
||||
nothing other than this License grants you permission to propagate or
|
||||
modify any covered work. These actions infringe copyright if you do
|
||||
not accept this License. Therefore, by modifying or propagating a
|
||||
covered work, you indicate your acceptance of this License to do so.
|
||||
|
||||
10. Automatic Licensing of Downstream Recipients.
|
||||
|
||||
Each time you convey a covered work, the recipient automatically
|
||||
receives a license from the original licensors, to run, modify and
|
||||
propagate that work, subject to this License. You are not responsible
|
||||
for enforcing compliance by third parties with this License.
|
||||
|
||||
An "entity transaction" is a transaction transferring control of an
|
||||
organization, or substantially all assets of one, or subdividing an
|
||||
organization, or merging organizations. If propagation of a covered
|
||||
work results from an entity transaction, each party to that
|
||||
transaction who receives a copy of the work also receives whatever
|
||||
licenses to the work the party's predecessor in interest had or could
|
||||
give under the previous paragraph, plus a right to possession of the
|
||||
Corresponding Source of the work from the predecessor in interest, if
|
||||
the predecessor has it or can get it with reasonable efforts.
|
||||
|
||||
You may not impose any further restrictions on the exercise of the
|
||||
rights granted or affirmed under this License. For example, you may
|
||||
not impose a license fee, royalty, or other charge for exercise of
|
||||
rights granted under this License, and you may not initiate litigation
|
||||
(including a cross-claim or counterclaim in a lawsuit) alleging that
|
||||
any patent claim is infringed by making, using, selling, offering for
|
||||
sale, or importing the Program or any portion of it.
|
||||
|
||||
11. Patents.
|
||||
|
||||
A "contributor" is a copyright holder who authorizes use under this
|
||||
License of the Program or a work on which the Program is based. The
|
||||
work thus licensed is called the contributor's "contributor version".
|
||||
|
||||
A contributor's "essential patent claims" are all patent claims
|
||||
owned or controlled by the contributor, whether already acquired or
|
||||
hereafter acquired, that would be infringed by some manner, permitted
|
||||
by this License, of making, using, or selling its contributor version,
|
||||
but do not include claims that would be infringed only as a
|
||||
consequence of further modification of the contributor version. For
|
||||
purposes of this definition, "control" includes the right to grant
|
||||
patent sublicenses in a manner consistent with the requirements of
|
||||
this License.
|
||||
|
||||
Each contributor grants you a non-exclusive, worldwide, royalty-free
|
||||
patent license under the contributor's essential patent claims, to
|
||||
make, use, sell, offer for sale, import and otherwise run, modify and
|
||||
propagate the contents of its contributor version.
|
||||
|
||||
In the following three paragraphs, a "patent license" is any express
|
||||
agreement or commitment, however denominated, not to enforce a patent
|
||||
(such as an express permission to practice a patent or covenant not to
|
||||
sue for patent infringement). To "grant" such a patent license to a
|
||||
party means to make such an agreement or commitment not to enforce a
|
||||
patent against the party.
|
||||
|
||||
If you convey a covered work, knowingly relying on a patent license,
|
||||
and the Corresponding Source of the work is not available for anyone
|
||||
to copy, free of charge and under the terms of this License, through a
|
||||
publicly available network server or other readily accessible means,
|
||||
then you must either (1) cause the Corresponding Source to be so
|
||||
available, or (2) arrange to deprive yourself of the benefit of the
|
||||
patent license for this particular work, or (3) arrange, in a manner
|
||||
consistent with the requirements of this License, to extend the patent
|
||||
license to downstream recipients. "Knowingly relying" means you have
|
||||
actual knowledge that, but for the patent license, your conveying the
|
||||
covered work in a country, or your recipient's use of the covered work
|
||||
in a country, would infringe one or more identifiable patents in that
|
||||
country that you have reason to believe are valid.
|
||||
|
||||
If, pursuant to or in connection with a single transaction or
|
||||
arrangement, you convey, or propagate by procuring conveyance of, a
|
||||
covered work, and grant a patent license to some of the parties
|
||||
receiving the covered work authorizing them to use, propagate, modify
|
||||
or convey a specific copy of the covered work, then the patent license
|
||||
you grant is automatically extended to all recipients of the covered
|
||||
work and works based on it.
|
||||
|
||||
A patent license is "discriminatory" if it does not include within
|
||||
the scope of its coverage, prohibits the exercise of, or is
|
||||
conditioned on the non-exercise of one or more of the rights that are
|
||||
specifically granted under this License. You may not convey a covered
|
||||
work if you are a party to an arrangement with a third party that is
|
||||
in the business of distributing software, under which you make payment
|
||||
to the third party based on the extent of your activity of conveying
|
||||
the work, and under which the third party grants, to any of the
|
||||
parties who would receive the covered work from you, a discriminatory
|
||||
patent license (a) in connection with copies of the covered work
|
||||
conveyed by you (or copies made from those copies), or (b) primarily
|
||||
for and in connection with specific products or compilations that
|
||||
contain the covered work, unless you entered into that arrangement,
|
||||
or that patent license was granted, prior to 28 March 2007.
|
||||
|
||||
Nothing in this License shall be construed as excluding or limiting
|
||||
any implied license or other defenses to infringement that may
|
||||
otherwise be available to you under applicable patent law.
|
||||
|
||||
12. No Surrender of Others' Freedom.
|
||||
|
||||
If conditions are imposed on you (whether by court order, agreement or
|
||||
otherwise) that contradict the conditions of this License, they do not
|
||||
excuse you from the conditions of this License. If you cannot convey a
|
||||
covered work so as to satisfy simultaneously your obligations under this
|
||||
License and any other pertinent obligations, then as a consequence you may
|
||||
not convey it at all. For example, if you agree to terms that obligate you
|
||||
to collect a royalty for further conveying from those to whom you convey
|
||||
the Program, the only way you could satisfy both those terms and this
|
||||
License would be to refrain entirely from conveying the Program.
|
||||
|
||||
13. Remote Network Interaction; Use with the GNU General Public License.
|
||||
|
||||
Notwithstanding any other provision of this License, if you modify the
|
||||
Program, your modified version must prominently offer all users
|
||||
interacting with it remotely through a computer network (if your version
|
||||
supports such interaction) an opportunity to receive the Corresponding
|
||||
Source of your version by providing access to the Corresponding Source
|
||||
from a network server at no charge, through some standard or customary
|
||||
means of facilitating copying of software. This Corresponding Source
|
||||
shall include the Corresponding Source for any work covered by version 3
|
||||
of the GNU General Public License that is incorporated pursuant to the
|
||||
following paragraph.
|
||||
|
||||
Notwithstanding any other provision of this License, you have
|
||||
permission to link or combine any covered work with a work licensed
|
||||
under version 3 of the GNU General Public License into a single
|
||||
combined work, and to convey the resulting work. The terms of this
|
||||
License will continue to apply to the part which is the covered work,
|
||||
but the work with which it is combined will remain governed by version
|
||||
3 of the GNU General Public License.
|
||||
|
||||
14. Revised Versions of this License.
|
||||
|
||||
The Free Software Foundation may publish revised and/or new versions of
|
||||
the GNU Affero General Public License from time to time. Such new versions
|
||||
will be similar in spirit to the present version, but may differ in detail to
|
||||
address new problems or concerns.
|
||||
|
||||
Each version is given a distinguishing version number. If the
|
||||
Program specifies that a certain numbered version of the GNU Affero General
|
||||
Public License "or any later version" applies to it, you have the
|
||||
option of following the terms and conditions either of that numbered
|
||||
version or of any later version published by the Free Software
|
||||
Foundation. If the Program does not specify a version number of the
|
||||
GNU Affero General Public License, you may choose any version ever published
|
||||
by the Free Software Foundation.
|
||||
|
||||
If the Program specifies that a proxy can decide which future
|
||||
versions of the GNU Affero General Public License can be used, that proxy's
|
||||
public statement of acceptance of a version permanently authorizes you
|
||||
to choose that version for the Program.
|
||||
|
||||
Later license versions may give you additional or different
|
||||
permissions. However, no additional obligations are imposed on any
|
||||
author or copyright holder as a result of your choosing to follow a
|
||||
later version.
|
||||
|
||||
15. Disclaimer of Warranty.
|
||||
|
||||
THERE IS NO WARRANTY FOR THE PROGRAM, TO THE EXTENT PERMITTED BY
|
||||
APPLICABLE LAW. EXCEPT WHEN OTHERWISE STATED IN WRITING THE COPYRIGHT
|
||||
HOLDERS AND/OR OTHER PARTIES PROVIDE THE PROGRAM "AS IS" WITHOUT WARRANTY
|
||||
OF ANY KIND, EITHER EXPRESSED OR IMPLIED, INCLUDING, BUT NOT LIMITED TO,
|
||||
THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
|
||||
PURPOSE. THE ENTIRE RISK AS TO THE QUALITY AND PERFORMANCE OF THE PROGRAM
|
||||
IS WITH YOU. SHOULD THE PROGRAM PROVE DEFECTIVE, YOU ASSUME THE COST OF
|
||||
ALL NECESSARY SERVICING, REPAIR OR CORRECTION.
|
||||
|
||||
16. Limitation of Liability.
|
||||
|
||||
IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING
|
||||
WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MODIFIES AND/OR CONVEYS
|
||||
THE PROGRAM AS PERMITTED ABOVE, BE LIABLE TO YOU FOR DAMAGES, INCLUDING ANY
|
||||
GENERAL, SPECIAL, INCIDENTAL OR CONSEQUENTIAL DAMAGES ARISING OUT OF THE
|
||||
USE OR INABILITY TO USE THE PROGRAM (INCLUDING BUT NOT LIMITED TO LOSS OF
|
||||
DATA OR DATA BEING RENDERED INACCURATE OR LOSSES SUSTAINED BY YOU OR THIRD
|
||||
PARTIES OR A FAILURE OF THE PROGRAM TO OPERATE WITH ANY OTHER PROGRAMS),
|
||||
EVEN IF SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE POSSIBILITY OF
|
||||
SUCH DAMAGES.
|
||||
|
||||
17. Interpretation of Sections 15 and 16.
|
||||
|
||||
If the disclaimer of warranty and limitation of liability provided
|
||||
above cannot be given local legal effect according to their terms,
|
||||
reviewing courts shall apply local law that most closely approximates
|
||||
an absolute waiver of all civil liability in connection with the
|
||||
Program, unless a warranty or assumption of liability accompanies a
|
||||
copy of the Program in return for a fee.
|
||||
|
||||
END OF TERMS AND CONDITIONS
|
||||
|
||||
How to Apply These Terms to Your New Programs
|
||||
|
||||
If you develop a new program, and you want it to be of the greatest
|
||||
possible use to the public, the best way to achieve this is to make it
|
||||
free software which everyone can redistribute and change under these terms.
|
||||
|
||||
To do so, attach the following notices to the program. It is safest
|
||||
to attach them to the start of each source file to most effectively
|
||||
state the exclusion of warranty; and each file should have at least
|
||||
the "copyright" line and a pointer to where the full notice is found.
|
||||
|
||||
<one line to give the program's name and a brief idea of what it does.>
|
||||
Copyright (C) <year> <name of author>
|
||||
|
||||
This program is free software: you can redistribute it and/or modify
|
||||
it under the terms of the GNU Affero General Public License as published
|
||||
by the Free Software Foundation, either version 3 of the License, or
|
||||
(at your option) any later version.
|
||||
|
||||
This program is distributed in the hope that it will be useful,
|
||||
but WITHOUT ANY WARRANTY; without even the implied warranty of
|
||||
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
||||
GNU Affero General Public License for more details.
|
||||
|
||||
You should have received a copy of the GNU Affero General Public License
|
||||
along with this program. If not, see <https://www.gnu.org/licenses/>.
|
||||
|
||||
Also add information on how to contact you by electronic and paper mail.
|
||||
|
||||
If your software can interact with users remotely through a computer
|
||||
network, you should also make sure that it provides a way for users to
|
||||
get its source. For example, if your program is a web application, its
|
||||
interface could display a "Source" link that leads users to an archive
|
||||
of the code. There are many ways you could offer source, and different
|
||||
solutions will be better for different programs; see section 13 for the
|
||||
specific requirements.
|
||||
|
||||
You should also get your employer (if you work as a programmer) or school,
|
||||
if any, to sign a "copyright disclaimer" for the program, if necessary.
|
||||
For more information on this, and how to apply and follow the GNU AGPL, see
|
||||
<https://www.gnu.org/licenses/>.
|
||||
@@ -1,269 +0,0 @@
|
||||
# Hermes slash-command monkeypatch
|
||||
|
||||
This document describes the Hermes Agent-side patch needed when `/bm-*` slash commands do not appear in the gateway even though the Basic Memory provider itself is installed and working.
|
||||
|
||||
This issue is tracked upstream in https://github.com/NousResearch/hermes-agent/issues/23603
|
||||
|
||||
Agents may use this file as a guide to add support for `/bm-*` slash commands in Hermes Agent builds that do not yet include the upstream fix.
|
||||
|
||||
## Current status and compatibility
|
||||
|
||||
This is a **Hermes Agent-side** compatibility patch, not a Basic Memory plugin runtime patch.
|
||||
|
||||
| Hermes Agent version | Basic Memory plugin version | Guidance |
|
||||
|---|---:|---|
|
||||
| `v0.13.x` | `v0.3.0` | Plugin `v0.3.0` remains the right runtime release. If `/bm-*` commands are missing, use the Hermes Agent-side monkeypatch below or wait for the upstream Hermes fix. |
|
||||
| `v0.14.0` / `v2026.5.16` | `v0.3.1` docs, runtime still equivalent to `v0.3.0` | The plugin runtime still works, but Hermes Agent `v0.14.0` still does **not** include the upstream slash-command discovery fix. Use the v0.14.0-compatible Hermes Agent-side patch below. |
|
||||
| Future Hermes release with upstream fix | Latest plugin | Do **not** apply this monkeypatch unless `/bm-*` commands are still absent; the fix should be redundant once Hermes loads active exclusive memory-provider commands during command discovery. |
|
||||
|
||||
Checked against Hermes Agent `v2026.5.16` / `v0.14.0` on 2026-05-16: the upstream Hermes release still does **not** include this fix. After applying the Hermes Agent-side patch below locally, `get_plugin_commands()` returns the expected `/bm-*` commands.
|
||||
|
||||
Important nuance: recent `hermes-basic-memory` versions include a best-effort PluginManager reach-in that registers commands when the provider is loaded. That workaround alone is not enough for gateway startup discovery in affected Hermes builds, because `get_plugin_commands()` does not load the active exclusive memory provider. The Hermes Agent-side patch is still needed until upstream command discovery loads the active memory provider and the memory-provider collector delegates command/skill registration.
|
||||
|
||||
Release/tagging note for agents: `v0.3.1` is a documentation release that clarifies Hermes Agent `v0.14.0` compatibility instructions. It does not require users on Hermes Agent `v0.13.x` to change plugin runtime behavior, and it should not be interpreted as a Basic Memory data/schema migration.
|
||||
|
||||
## Problem
|
||||
|
||||
`hermes-basic-memory` is an **exclusive memory-provider plugin**. Hermes loads exclusive memory providers through `plugins.memory`, not through the normal `PluginManager` discovery path.
|
||||
|
||||
Gateway adapters register native slash commands during startup by calling Hermes's plugin command discovery APIs. In affected Hermes builds, that startup path only sees commands registered by normal plugins. The active memory provider has not been loaded yet, and the memory-provider loader uses a collector that captures only `register_memory_provider(...)`. As a result, commands registered by this plugin with `ctx.register_command(...)` never reach the central plugin command registry before Discord/native slash-command sync.
|
||||
|
||||
Symptoms:
|
||||
|
||||
- `hermes memory status` shows `Provider: basic-memory` and `Status: available`.
|
||||
- Agent tools such as `bm_search`, `bm_read`, and `bm_recent` work.
|
||||
- Native slash commands such as `/bm-search`, `/bm-read`, and `/bm-context` are missing after `hermes gateway restart`.
|
||||
|
||||
## Target behavior
|
||||
|
||||
When Hermes builds its plugin command list, it should also load the configured active memory provider once, allowing that provider to register commands and skills into the same central registries used by ordinary plugins.
|
||||
|
||||
After the patch, `get_plugin_commands()` should include commands such as:
|
||||
|
||||
```text
|
||||
bm-context
|
||||
bm-project
|
||||
bm-read
|
||||
bm-recent
|
||||
bm-remember
|
||||
bm-search
|
||||
bm-status
|
||||
bm-workspace
|
||||
```
|
||||
|
||||
## Files to patch in Hermes Agent
|
||||
|
||||
Patch these files in the Hermes Agent repository, not in this plugin repository:
|
||||
|
||||
```text
|
||||
hermes_cli/plugins.py
|
||||
plugins/memory/__init__.py
|
||||
```
|
||||
|
||||
Recommended tests to add/update in Hermes Agent:
|
||||
|
||||
```text
|
||||
tests/hermes_cli/test_plugin_cli_registration.py
|
||||
tests/hermes_cli/test_plugins.py
|
||||
```
|
||||
|
||||
## Implementation outline
|
||||
|
||||
### 1. Load active memory-provider commands from `get_plugin_commands()`
|
||||
|
||||
In `hermes_cli/plugins.py`, add module-level idempotency/recursion guards near the global plugin manager:
|
||||
|
||||
```python
|
||||
_plugin_manager: Optional[PluginManager] = None
|
||||
_memory_provider_command_loads: set[str] = set()
|
||||
_memory_provider_command_loading = False
|
||||
```
|
||||
|
||||
Update `get_plugin_commands()` so it first ensures normal plugin discovery, then best-effort loads the active memory provider before returning the command registry:
|
||||
|
||||
```python
|
||||
def get_plugin_commands() -> Dict[str, dict]:
|
||||
"""Return the full plugin commands dict (name → {handler, description, plugin}).
|
||||
|
||||
Triggers idempotent plugin discovery so callers can use plugin commands
|
||||
before any explicit discover_plugins() call. Also initializes the active
|
||||
memory provider once so exclusive memory-provider plugins can contribute
|
||||
gateway slash commands during startup discovery.
|
||||
"""
|
||||
manager = _ensure_plugins_discovered()
|
||||
_ensure_active_memory_provider_commands_loaded()
|
||||
return manager._plugin_commands
|
||||
```
|
||||
|
||||
Add the helper:
|
||||
|
||||
```python
|
||||
def _ensure_active_memory_provider_commands_loaded() -> None:
|
||||
"""Best-effort load of the active memory provider's slash commands."""
|
||||
global _memory_provider_command_loading
|
||||
if _memory_provider_command_loading:
|
||||
return
|
||||
try:
|
||||
from plugins import memory as memory_plugins
|
||||
|
||||
active = memory_plugins._get_active_memory_provider()
|
||||
if not active or active in _memory_provider_command_loads:
|
||||
return
|
||||
_memory_provider_command_loading = True
|
||||
try:
|
||||
memory_plugins.load_memory_provider(active)
|
||||
_memory_provider_command_loads.add(active)
|
||||
finally:
|
||||
_memory_provider_command_loading = False
|
||||
except Exception as exc:
|
||||
logger.debug(
|
||||
"Failed to load active memory-provider plugin commands: %s",
|
||||
exc,
|
||||
exc_info=_PLUGINS_DEBUG,
|
||||
)
|
||||
```
|
||||
|
||||
Notes:
|
||||
|
||||
- This must be best-effort; command discovery should not break Hermes startup if a memory provider is misconfigured.
|
||||
- The recursion guard prevents `load_memory_provider(...)` → provider `register(...)` → `ctx.register_command(...)` → plugin manager access from re-entering endlessly.
|
||||
- The load set prevents duplicate provider command registration work.
|
||||
|
||||
### 2. Make the memory-provider collector delegate commands and skills
|
||||
|
||||
In `plugins/memory/__init__.py`, import `Callable`:
|
||||
|
||||
```python
|
||||
from typing import Callable, List, Optional, Tuple
|
||||
```
|
||||
|
||||
When loading a provider directory, pass the plugin/provider name to the collector:
|
||||
|
||||
```python
|
||||
collector = _ProviderCollector(plugin_name=name)
|
||||
```
|
||||
|
||||
Replace the collector that only captures `register_memory_provider(...)` with a plugin-context shim that also delegates `register_command(...)` and `register_skill(...)` into the central `PluginManager` registries:
|
||||
|
||||
```python
|
||||
class _ProviderCollector:
|
||||
"""Plugin-context shim used while loading memory providers.
|
||||
|
||||
Memory providers are exclusive plugins and are loaded by this module
|
||||
instead of the general PluginManager. They still need access to the same
|
||||
slash-command and skill registries as normal plugins, otherwise active
|
||||
memory-provider commands are invisible during gateway startup discovery.
|
||||
"""
|
||||
|
||||
def __init__(self, plugin_name: str = "memory-provider"):
|
||||
self.provider = None
|
||||
self.plugin_name = plugin_name
|
||||
|
||||
def register_memory_provider(self, provider):
|
||||
self.provider = provider
|
||||
|
||||
def register_command(
|
||||
self,
|
||||
name: str,
|
||||
handler: Callable,
|
||||
description: str = "",
|
||||
args_hint: str = "",
|
||||
) -> None:
|
||||
"""Register a memory-provider slash command with PluginManager."""
|
||||
try:
|
||||
from hermes_cli.plugins import _ensure_plugins_discovered
|
||||
except Exception:
|
||||
return
|
||||
|
||||
clean = name.lower().strip().lstrip("/").replace(" ", "-")
|
||||
if not clean:
|
||||
return
|
||||
|
||||
try:
|
||||
manager = _ensure_plugins_discovered()
|
||||
except Exception:
|
||||
return
|
||||
|
||||
plugin_commands = getattr(manager, "_plugin_commands", None)
|
||||
if plugin_commands is None:
|
||||
return
|
||||
plugin_commands[clean] = {
|
||||
"handler": handler,
|
||||
"description": description or "Plugin command",
|
||||
"plugin": self.plugin_name,
|
||||
"args_hint": (args_hint or "").strip(),
|
||||
}
|
||||
|
||||
def register_skill(
|
||||
self,
|
||||
name: str,
|
||||
path: Path,
|
||||
description: str = "",
|
||||
) -> None:
|
||||
"""Register a memory-provider skill with PluginManager."""
|
||||
try:
|
||||
from agent.skill_utils import _NAMESPACE_RE
|
||||
from hermes_cli.plugins import _ensure_plugins_discovered
|
||||
except Exception:
|
||||
return
|
||||
|
||||
if ":" in name or not name or not _NAMESPACE_RE.match(name):
|
||||
raise ValueError(f"Invalid skill name '{name}'.")
|
||||
if not path.exists():
|
||||
raise FileNotFoundError(f"SKILL.md not found at {path}")
|
||||
|
||||
try:
|
||||
manager = _ensure_plugins_discovered()
|
||||
except Exception:
|
||||
return
|
||||
|
||||
plugin_skills = getattr(manager, "_plugin_skills", None)
|
||||
if plugin_skills is None:
|
||||
return
|
||||
plugin_skills[f"{self.plugin_name}:{name}"] = {
|
||||
"path": path,
|
||||
"plugin": self.plugin_name,
|
||||
"bare_name": name,
|
||||
"description": description,
|
||||
}
|
||||
```
|
||||
|
||||
Keep existing no-op methods such as `register_tool(...)` and `register_cli_command(...)` as no-ops unless the target Hermes version expects otherwise.
|
||||
|
||||
## Verification
|
||||
|
||||
From the Hermes Agent repository, run focused compile/tests:
|
||||
|
||||
```bash
|
||||
python -m py_compile hermes_cli/plugins.py plugins/memory/__init__.py
|
||||
python -m pytest \
|
||||
tests/hermes_cli/test_plugins.py::TestPluginCommands::test_get_plugin_commands_loads_active_memory_provider_commands \
|
||||
tests/hermes_cli/test_plugin_cli_registration.py::TestProviderCollectorRegistration \
|
||||
-q -o 'addopts='
|
||||
```
|
||||
|
||||
Then verify the active config sees the Basic Memory commands:
|
||||
|
||||
```bash
|
||||
python - <<'PY'
|
||||
import hermes_cli.plugins as p
|
||||
p._plugin_manager = None
|
||||
p._memory_provider_command_loads.clear()
|
||||
cmds = p.get_plugin_commands()
|
||||
print(sorted(k for k in cmds if k.startswith('bm-')))
|
||||
PY
|
||||
```
|
||||
|
||||
Expected output:
|
||||
|
||||
```text
|
||||
['bm-context', 'bm-project', 'bm-read', 'bm-recent', 'bm-remember', 'bm-search', 'bm-status', 'bm-workspace']
|
||||
```
|
||||
|
||||
Finally restart the gateway so native slash commands are synced:
|
||||
|
||||
```bash
|
||||
hermes gateway restart
|
||||
```
|
||||
|
||||
For Discord, global command propagation can lag briefly. If the commands do not show immediately, type `/bm` directly or reload the Discord client.
|
||||
@@ -1,235 +0,0 @@
|
||||
# hermes-basic-memory
|
||||
|
||||
[](https://www.gnu.org/licenses/agpl-3.0)
|
||||
|
||||
Hermes Memory Provider plugin that gives [Hermes Agent](https://github.com/NousResearch/hermes-agent) a persistent knowledge graph backed by [Basic Memory](https://github.com/basicmachines-co/basic-memory).
|
||||
|
||||
The plugin replaces Hermes's "no external memory provider" with a real graph: search-before-answer recall, per-turn capture, end-of-session summaries, and ten `bm_*` tools the agent can call directly. Local mode by default; one CLI flip switches to true cloud routing through Basic Memory Cloud.
|
||||
|
||||
## Install
|
||||
|
||||
```bash
|
||||
hermes plugins install basicmachines-co/basic-memory --path integrations/hermes
|
||||
```
|
||||
|
||||
Then activate it in `~/.hermes/config.yaml`:
|
||||
|
||||
```yaml
|
||||
memory:
|
||||
provider: basic-memory
|
||||
```
|
||||
|
||||
If you run the gateway, restart it (`hermes gateway restart`). Done.
|
||||
|
||||
If your installed Hermes build does not support `--path`, use the final deprecated `basicmachines-co/hermes-basic-memory` pointer release until Hermes subpath installs are available. Ongoing development now lives in [`basic-memory/integrations/hermes`](https://github.com/basicmachines-co/basic-memory/tree/main/integrations/hermes).
|
||||
|
||||
The plugin self-installs the `basic-memory` CLI on first init via `uv tool install basic-memory` (one-time ~10s pause if it isn't already present). The bm binary lands at `~/.local/bin/bm` — the same location a manual `uv tool install basic-memory` would produce, so a later manual install or upgrade is a no-op rather than a second install.
|
||||
|
||||
### Prerequisites
|
||||
|
||||
- [Hermes Agent](https://github.com/NousResearch/hermes-agent)
|
||||
- [`uv`](https://docs.astral.sh/uv/) on PATH (used for the bootstrap install)
|
||||
- The `mcp` Python package in the Hermes venv. If `hermes plugins install` doesn't auto-install it (it follows `pip_dependencies` in `plugin.yaml`), run:
|
||||
```bash
|
||||
uv pip install --python ~/.hermes/hermes-agent/venv/bin/python mcp
|
||||
```
|
||||
|
||||
### Verify
|
||||
|
||||
```bash
|
||||
hermes memory status
|
||||
```
|
||||
|
||||
Expected:
|
||||
```
|
||||
Provider: basic-memory
|
||||
Plugin: installed ✓
|
||||
Status: available ✓
|
||||
```
|
||||
|
||||
## What the agent gets
|
||||
|
||||
Ten tools (curated subset of Basic Memory's MCP surface):
|
||||
|
||||
| Tool | Use |
|
||||
|---|---|
|
||||
| `bm_search` | Semantic + full-text search; **call this before answering** |
|
||||
| `bm_read` | Fetch a note by title, permalink, or `memory://` URL |
|
||||
| `bm_write` | Create a new note (capture decisions, meeting notes, insights) |
|
||||
| `bm_edit` | Append, prepend, find/replace, replace-section |
|
||||
| `bm_context` | Navigate via `memory://` URLs to find related notes |
|
||||
| `bm_delete` | Delete a note |
|
||||
| `bm_move` | Move a note to a different folder |
|
||||
| `bm_recent` | List notes updated recently (default `7d`; accepts natural-language timeframes) |
|
||||
| `bm_projects` | List available projects with their UUIDs (for cross-project routing) |
|
||||
| `bm_workspaces` | List Basic Memory Cloud workspaces |
|
||||
|
||||
Every read/write tool also accepts optional `project` / `project_id` for per-call routing — write or read against a project other than the configured one without reconfiguring the plugin.
|
||||
|
||||
Plus automatic capture:
|
||||
- **Per turn**: every user/assistant exchange appends to a running session-transcript note
|
||||
- **End of session**: a separate summary note is written, linked back to the transcript via a `summary_of` relation
|
||||
|
||||
A bundled skill (`skill:view basic-memory:basic-memory`) gives the agent a longer reference doc on top of the always-on `system_prompt_block`.
|
||||
|
||||
## Slash commands
|
||||
|
||||
For direct, in-session use without going through the agent (requires Hermes ≥ v0.11.0):
|
||||
|
||||
| Command | Use |
|
||||
|---|---|
|
||||
| `/bm-search <query>` | Search the knowledge graph; returns compact title/permalink/preview rows. |
|
||||
| `/bm-read <identifier>` | Read a note by title, permalink, or `memory://` URL. |
|
||||
| `/bm-context <identifier>` | Build context for a note (target + related). |
|
||||
| `/bm-recent [timeframe]` | Recently updated notes. Default `7d`; accepts `"2 weeks"`, `"yesterday"`, etc. |
|
||||
| `/bm-status` | Plugin/provider state: mode, project, capture flags, bm CLI path. |
|
||||
| `/bm-remember <text>` | Capture a quick note. Title = first line (≤80 chars), folder = `remember_folder` (default `bm-remember`), tagged `manual-capture`. |
|
||||
| `/bm-project` | List all known projects; the active one is marked. |
|
||||
| `/bm-workspace` | List BM Cloud workspaces. Cloud mode only — prints an explanatory line in local mode. |
|
||||
|
||||
Examples:
|
||||
|
||||
```text
|
||||
/bm-search Q3 OKRs
|
||||
/bm-read decisions/auth-rewrite
|
||||
/bm-recent yesterday
|
||||
/bm-remember Reminder: switch the staging job to the new image after the rebase lands.
|
||||
```
|
||||
|
||||
`/bm-project` and `/bm-workspace` are read-only in 0.2.0 — mid-session switching is intentionally not supported because auto-capture would otherwise land in the wrong place. Tracked as a follow-up.
|
||||
|
||||
### Known issue: `/bm-*` commands may not appear in some Hermes gateway builds
|
||||
|
||||
Plugin v0.2.0 and later register the commands above, but some Hermes Agent gateway builds do not discover slash commands contributed by an **exclusive memory-provider plugin** during startup. The symptoms are:
|
||||
|
||||
- the memory tools work for the agent (`bm_search`, `bm_read`, etc.);
|
||||
- `hermes memory status` shows `Provider: basic-memory` and `Status: available`; but
|
||||
- Discord/native slash command pickers do not show `/bm-search`, `/bm-read`, `/bm-context`, and the other `/bm-*` commands after `hermes gateway restart`.
|
||||
|
||||
This is a Hermes Agent plugin-discovery issue, not a Basic Memory runtime issue. It is tracked upstream in [NousResearch/hermes-agent#23603](https://github.com/NousResearch/hermes-agent/issues/23603). Updating the Basic Memory plugin alone cannot fix affected gateway startup discovery; Hermes Agent itself must include or receive the compatibility patch. Until the upstream Hermes fix is available in your installed Hermes version, use one of these workarounds:
|
||||
|
||||
1. apply the Hermes Agent-side patch described in [MONKEYPATCH.md](MONKEYPATCH.md), which includes compatibility notes for Hermes Agent v0.13.x and v0.14.0; or
|
||||
2. use the agent tools directly (`bm_search`, `bm_read`, `bm_recent`, etc.) instead of native slash commands.
|
||||
|
||||
After applying an updated or patched Hermes build, restart the gateway so Discord/native slash commands are re-synced:
|
||||
|
||||
```bash
|
||||
hermes gateway restart
|
||||
```
|
||||
|
||||
If Discord still does not show the commands immediately, type `/bm` directly or reload the Discord client; global command propagation can lag briefly.
|
||||
|
||||
## Configuration
|
||||
|
||||
Defaults are reasonable for local use:
|
||||
|
||||
| Key | Default | Notes |
|
||||
|---|---|---|
|
||||
| `mode` | `local` | `local` (in-process) or `cloud` (route through BM Cloud API) |
|
||||
| `project` | `hermes-memory` | BM project name |
|
||||
| `project_path` | `~/hermes-memory/` | Local mode only — where session notes land |
|
||||
| `capture_folder` | `hermes-sessions` | Folder within the project for session notes |
|
||||
| `capture_per_turn` | `true` | Append every turn to a session transcript |
|
||||
| `capture_session_end` | `true` | Write a summary note when the session ends |
|
||||
| `remember_folder` | `bm-remember` | Folder where `/bm-remember` captures land (kept separate from session transcripts) |
|
||||
|
||||
To override, write `~/.hermes/basic-memory.json` or run `hermes memory setup basic-memory`:
|
||||
|
||||
```json
|
||||
{
|
||||
"mode": "local",
|
||||
"project": "hermes-memory",
|
||||
"project_path": "~/hermes-memory/",
|
||||
"capture_per_turn": true,
|
||||
"capture_session_end": true,
|
||||
"capture_folder": "hermes-sessions",
|
||||
"remember_folder": "bm-remember"
|
||||
}
|
||||
```
|
||||
|
||||
In local mode the plugin auto-creates the BM project on first init via `bm project add`. In cloud mode it doesn't — you create the cloud-routed project yourself (see below) and the plugin verifies it's registered before initializing.
|
||||
|
||||
### Cloud mode
|
||||
|
||||
When `mode: cloud`, tool calls route directly through the BM cloud API — no local file mirror, no bisync. You set this up once with the BM CLI:
|
||||
|
||||
```bash
|
||||
# Authenticate (OAuth) or save an API key
|
||||
bm cloud login # OAuth — interactive
|
||||
# OR for headless/automation:
|
||||
bm cloud create-key "hermes"
|
||||
bm cloud set-key bmc_...
|
||||
|
||||
# Create the project, then flip it to cloud routing.
|
||||
# --workspace is required if you belong to more than one workspace
|
||||
# (otherwise BM auto-resolves the only one available).
|
||||
bm project add hermes-memory-cloud
|
||||
bm project set-cloud hermes-memory-cloud --workspace Personal
|
||||
|
||||
# Point the plugin at it
|
||||
cat > ~/.hermes/basic-memory.json <<EOF
|
||||
{
|
||||
"mode": "cloud",
|
||||
"project": "hermes-memory-cloud",
|
||||
"capture_per_turn": true,
|
||||
"capture_session_end": true,
|
||||
"capture_folder": "hermes-sessions"
|
||||
}
|
||||
EOF
|
||||
|
||||
hermes gateway restart
|
||||
```
|
||||
|
||||
Tool calls now route from `bm mcp` → `<cloud_host>/proxy` over HTTPS using your OAuth token (or API key). Notes never touch local disk.
|
||||
|
||||
**Don't confuse cloud mode with `bm cloud bisync`.** Bisync is rclone-style two-way file sync between a *local* project and cloud storage, intended for keeping local working copies. For agent-driven capture you want true cloud routing (`set-cloud`), not bisync.
|
||||
|
||||
## Updating / removing
|
||||
|
||||
```bash
|
||||
hermes plugins update basic-memory
|
||||
hermes plugins remove basic-memory # then revert memory.provider in config.yaml
|
||||
```
|
||||
|
||||
## Foot-guns
|
||||
|
||||
- **`<memory-context>` tags in notes**: Hermes's streaming output scrubber strips literal `<memory-context>...</memory-context>` blocks from assistant text. If a note contains those tags and the assistant echoes the body verbatim, the echoed copy gets eaten mid-stream. Tool results inbound are unaffected. Avoid those tags in BM notes; if you must include them, fence in a code block.
|
||||
- **Single external provider**: Hermes accepts only one external memory provider at a time. Activating basic-memory displaces any other.
|
||||
- **CLI cold start**: `hermes -z ...` invocations spawn `bm mcp` per run (~2-5s). Long-running gateway sessions amortize this.
|
||||
- **Multiple cloud workspaces**: if your BM Cloud account belongs to more than one workspace, `bm project set-cloud` must be invoked with `--workspace <name>`. Otherwise tool calls fail with "Multiple workspaces are available".
|
||||
|
||||
## Development
|
||||
|
||||
The plugin is a single-file Python module at `__init__.py`. The Hermes plugin loader expects `register(ctx)` and grep-detects either `register_memory_provider` or `MemoryProvider` in the file.
|
||||
|
||||
For local development (point Hermes at your working tree instead of going through `hermes plugins install`):
|
||||
|
||||
```bash
|
||||
git clone https://github.com/basicmachines-co/basic-memory ~/code/basic-memory
|
||||
mkdir -p ~/.hermes/plugins
|
||||
ln -snf ~/code/basic-memory/integrations/hermes ~/.hermes/plugins/basic-memory
|
||||
```
|
||||
|
||||
### Running tests
|
||||
|
||||
```bash
|
||||
# From the monorepo root
|
||||
just package-check-hermes
|
||||
|
||||
# Or from integrations/hermes
|
||||
just check
|
||||
|
||||
# Unit tests (fast, hermetic — no Hermes or bm required)
|
||||
uv run --with pytest pytest
|
||||
|
||||
# Integration tests (gated — exercise every tool against a real bm MCP server)
|
||||
BM_INTEGRATION=1 uv run --with pytest --with mcp pytest tests/test_integration.py
|
||||
```
|
||||
|
||||
The unit suite stubs out Hermes-internal imports (`agent.memory_provider`, `tools.registry`) so it runs without a Hermes install. `mcp` is optional at unit-test time — its absence just makes `is_available()` return False, which the tests verify.
|
||||
|
||||
Integration tests require `BM_INTEGRATION=1`, `bm` CLI on PATH, and `mcp` Python package importable. Each session creates a unique throwaway BM project (under `tempfile.mkdtemp`) and removes it on teardown, so they never touch your real BM projects.
|
||||
|
||||
## License
|
||||
|
||||
AGPL-3.0-or-later, matching [basic-memory](https://github.com/basicmachines-co/basic-memory). See [LICENSE](LICENSE).
|
||||
File diff suppressed because it is too large
Load Diff
@@ -1,22 +0,0 @@
|
||||
# Basic Memory Hermes plugin checks
|
||||
|
||||
repo_root := "../.."
|
||||
|
||||
# Validate plugin.yaml, module entrypoint, bundled skill, and test layout.
|
||||
manifest-check:
|
||||
python3 {{repo_root}}/scripts/validate_hermes_plugin.py .
|
||||
|
||||
# Unit tests are hermetic and do not require a Hermes install.
|
||||
test:
|
||||
uv run --no-project --with pytest --with pytest-cov --python 3.12 pytest -q --cov=. --cov-report=term-missing
|
||||
|
||||
# Gated integration test against a real bm MCP server.
|
||||
test-int:
|
||||
BM_INTEGRATION=1 uv run --no-project --with pytest --with mcp --python 3.12 pytest tests/test_integration.py -q
|
||||
|
||||
# Full local check.
|
||||
check: manifest-check test
|
||||
|
||||
# Show available recipes
|
||||
default:
|
||||
@just --list
|
||||
@@ -1,11 +0,0 @@
|
||||
name: basic-memory
|
||||
version: 0.22.1
|
||||
description: "Basic Memory — persistent knowledge graph backed by the basic-memory MCP server"
|
||||
pip_dependencies:
|
||||
- mcp
|
||||
hooks:
|
||||
- prefetch
|
||||
- queue_prefetch
|
||||
- sync_turn
|
||||
- on_session_end
|
||||
- shutdown
|
||||
@@ -1,8 +0,0 @@
|
||||
[pytest]
|
||||
testpaths = tests
|
||||
addopts = -ra
|
||||
pythonpath =
|
||||
.
|
||||
tests/stubs
|
||||
filterwarnings =
|
||||
ignore::DeprecationWarning
|
||||
@@ -1 +0,0 @@
|
||||
pytest>=7.0
|
||||
@@ -1,242 +0,0 @@
|
||||
---
|
||||
name: basic-memory
|
||||
description: Use the Basic Memory knowledge graph for persistent memory across sessions. Search before answering; capture decisions, meetings, and insights as notes.
|
||||
category: memory
|
||||
---
|
||||
|
||||
# Basic Memory Knowledge Graph
|
||||
|
||||
You have access to a persistent knowledge graph backed by Basic Memory. The graph survives across sessions and is shared with other tools (Claude Desktop, Obsidian, the `bm` CLI). Use the `bm_*` tools below to recall and capture information.
|
||||
|
||||
## Use `bm_*`, not the `bm` CLI
|
||||
|
||||
**Always invoke the `bm_*` tools directly. Do not shell out to the `bm` CLI for note operations.**
|
||||
|
||||
The `bm_*` tools route through a persistent MCP connection — roughly 0.1 seconds per call. Running `bm` from the shell spawns a fresh Python process per call (1-2 seconds of cold-start every time) and bypasses Hermes's automatic per-turn capture, so the session-transcript and summary notes won't reflect what you did.
|
||||
|
||||
The CLI is fine when you genuinely need a feature these wrappers don't expose (rare). Otherwise, prefer:
|
||||
|
||||
| Use case | Tool (not CLI) |
|
||||
|---|---|
|
||||
| Search the graph | `bm_search` |
|
||||
| Read a note | `bm_read` |
|
||||
| Create / update a note | `bm_write` / `bm_edit` |
|
||||
| Navigate relations | `bm_context` |
|
||||
| Move / delete | `bm_move` / `bm_delete` |
|
||||
| What's been touched lately | `bm_recent` |
|
||||
| List available projects | `bm_projects` |
|
||||
| List cloud workspaces | `bm_workspaces` |
|
||||
|
||||
## Tool reference
|
||||
|
||||
### `bm_search` — search the graph
|
||||
Use **before** answering questions about prior decisions, projects, meetings, or anything that might already be documented.
|
||||
|
||||
```
|
||||
bm_search({ query: "auth strategy decision", limit: 5 })
|
||||
```
|
||||
|
||||
### `bm_read` — fetch a note's full content
|
||||
After search shows a relevant note, read it for context.
|
||||
|
||||
```
|
||||
bm_read({ identifier: "decisions/auth-strategy" })
|
||||
bm_read({ identifier: "memory://projects/api-redesign" })
|
||||
```
|
||||
|
||||
### `bm_context` — navigate via memory:// URLs
|
||||
Returns the target note plus related notes via traversed relations.
|
||||
|
||||
```
|
||||
bm_context({ url: "memory://projects/api-redesign", depth: 1 })
|
||||
```
|
||||
|
||||
### `bm_write` — capture new knowledge
|
||||
When the user shares a decision, meeting outcome, or insight worth keeping, capture it. Use clear titles and a folder.
|
||||
|
||||
```
|
||||
bm_write({
|
||||
title: "API Authentication Decision",
|
||||
folder: "decisions",
|
||||
content: "# API Authentication\n\n## Context\n...\n\n## Decision\n..."
|
||||
})
|
||||
```
|
||||
|
||||
Recommended folders: `projects/`, `decisions/`, `meetings/`, `concepts/`, `weekly/`.
|
||||
|
||||
### `bm_edit` — incremental updates
|
||||
Operations: `append`, `prepend`, `find_replace` (requires `find_text`), `replace_section` (requires `section`).
|
||||
|
||||
```
|
||||
bm_edit({
|
||||
identifier: "projects/api-redesign",
|
||||
operation: "append",
|
||||
content: "\n## Update 2026-05-09\nDeployed to staging."
|
||||
})
|
||||
```
|
||||
|
||||
### `bm_delete` / `bm_move` — maintenance
|
||||
Use sparingly. `bm_move` takes `new_folder`.
|
||||
|
||||
### `bm_recent` — what's been touched lately
|
||||
Returns notes updated within a window. Use when there's no specific query yet — e.g. "what was I working on yesterday?"
|
||||
|
||||
```
|
||||
bm_recent({ timeframe: "7d" })
|
||||
bm_recent({ timeframe: "yesterday", limit: 20 })
|
||||
bm_recent({ timeframe: "2 weeks", type: "entity" })
|
||||
```
|
||||
|
||||
`timeframe` accepts natural language (`"yesterday"`, `"2 weeks"`, `"last month"`) or compact forms (`"7d"`, `"24h"`). Default is `7d`.
|
||||
|
||||
### `bm_projects` — list available projects
|
||||
Returns name, workspace slug, and `external_id` (UUID) per project across local and cloud. Call this when the user names a project that isn't the active one. Route follow-up tool calls either by workspace-qualified name (`project: "personal/main"`) or by UUID (`project_id: "bf2a4c1e-d77f-..."`) — see Cross-project routing below.
|
||||
|
||||
```
|
||||
bm_projects()
|
||||
```
|
||||
|
||||
### `bm_workspaces` — list BM Cloud workspaces
|
||||
Workspaces are a BM Cloud concept. Returns name, type, role, and default flag. Pair with `bm_projects` when the same project name might exist in more than one workspace and you need to disambiguate.
|
||||
|
||||
```
|
||||
bm_workspaces()
|
||||
```
|
||||
|
||||
## Permalinks
|
||||
|
||||
A permalink is the canonical, URL-friendly identifier for a note. Three shapes exist; the read/write tools accept all of them:
|
||||
|
||||
| Shape | Example | When |
|
||||
|---|---|---|
|
||||
| **Short** | `decisions/auth-strategy` | Bare `folder/note-slug`. Tools need a `project` (or `project_id`) arg to route — the permalink alone isn't enough. |
|
||||
| **Project-qualified** | `main/decisions/auth-strategy` | `project-name/folder/note-slug`. Carries enough context to route without a separate `project` arg. |
|
||||
| **Workspace-qualified** | `personal/main/decisions/auth-strategy` | `workspace-slug/project-name/folder/note-slug`. Fully routes, including across cloud workspaces with same-named projects. |
|
||||
|
||||
**Important: the permalink returned by `bm_write` already encodes the routing it needs for follow-up reads.** If you wrote with `project="personal/main"`, you get back `personal/main/folder/note-slug` and can call `bm_read({ identifier: <that permalink> })` with no `project` arg. The permalink self-routes.
|
||||
|
||||
`memory://` URLs follow the same shapes: `memory://personal/main/decisions/auth-strategy` is valid. The `memory://` prefix is optional for `bm_read` (any of the three permalink shapes works directly); `bm_context` expects the prefix.
|
||||
|
||||
## Cross-project routing
|
||||
|
||||
Every read/write tool (`bm_search`, `bm_read`, `bm_write`, `bm_edit`, `bm_context`, `bm_delete`, `bm_move`, `bm_recent`) accepts optional `project` and `project_id`:
|
||||
|
||||
- `project` — project name, optionally workspace-qualified. Plain (`"main"`) when the name is globally unique; qualified (`"personal/main"`, `"team-paul/research"`) when you need to pick a specific cloud workspace by slug.
|
||||
- `project_id` — UUID from `bm_projects` (`external_id` field). The most stable identifier — survives project renames and works across workspaces without qualification. Wins over `project` if both are passed.
|
||||
|
||||
Omit both and the call uses the Hermes-configured active project.
|
||||
|
||||
```
|
||||
# Plain project name (unique)
|
||||
bm_write({ title: "...", folder: "...", content: "...", project: "main" })
|
||||
|
||||
# Workspace-qualified name (disambiguates same-named projects across workspaces)
|
||||
bm_write({ title: "...", folder: "...", content: "...", project: "personal/main" })
|
||||
|
||||
# UUID (most stable, survives renames)
|
||||
bm_write({ title: "...", folder: "...", content: "...", project_id: "bf2a4c1e-d77f-..." })
|
||||
```
|
||||
|
||||
`bm_projects` and `bm_workspaces` themselves do **not** take routing — they list across everything.
|
||||
|
||||
## Recipe: writing an existing file into a specific project
|
||||
|
||||
When the user asks something like *"save this markdown file to my personal `main` project, return the permalink"*:
|
||||
|
||||
1. **Discover the project.** Call `bm_projects()` and find the entry matching the user's described project + workspace. You can route by either the workspace-qualified name (`personal/main`) or the UUID (`external_id`).
|
||||
|
||||
```
|
||||
bm_projects()
|
||||
# → [{name: "main", external_id: "bf2a4c1e-d77f-4b7a-9c3e-5d8a1f0e2b6d", workspace: "Personal", ...}, ...]
|
||||
```
|
||||
|
||||
If a project name appears in multiple workspaces, use `bm_workspaces()` to confirm which slug you want.
|
||||
|
||||
2. **Read the file from disk.** Use Hermes's filesystem tool (not a `bm_*` tool — local files aren't in the graph yet).
|
||||
|
||||
3. **Write the note with explicit routing.** Either form works; the workspace-qualified name reads cleaner in logs, the UUID is more durable.
|
||||
|
||||
```
|
||||
bm_write({
|
||||
title: "StartWithDrew Level 9 Task Queue",
|
||||
folder: "startwithdrew",
|
||||
content: <file body>,
|
||||
project: "personal/main"
|
||||
})
|
||||
# → returns "personal/main/startwithdrew/start-with-drew-level-9-task-queue"
|
||||
# (the returned permalink is workspace-qualified — carries its own routing)
|
||||
```
|
||||
|
||||
4. **Verify by reading back.** No `project` arg needed — the workspace-qualified permalink routes itself.
|
||||
|
||||
```
|
||||
bm_read({ identifier: "personal/main/startwithdrew/start-with-drew-level-9-task-queue" })
|
||||
```
|
||||
|
||||
Return the permalink (and the project name for clarity) to the user.
|
||||
|
||||
## When to use each tool
|
||||
|
||||
| Situation | Tool |
|
||||
|---|---|
|
||||
| User asks about a topic that might already be documented | `bm_search` first, then `bm_read` |
|
||||
| User exposes a decision, plan, or meeting outcome | offer to `bm_write` |
|
||||
| Updating prior work | `bm_edit` (append for time-ordered logs, replace_section for living docs) |
|
||||
| Exploring related concepts | `bm_context` |
|
||||
| "What was I working on yesterday?" / no specific query yet | `bm_recent` |
|
||||
| User names a project that isn't the active one | `bm_projects` → call read/write tool with `project: "workspace/name"` or `project_id: "<uuid>"` |
|
||||
| Same project name might exist in multiple workspaces | `bm_projects` (+ `bm_workspaces` if needed) → route with workspace-qualified `project` or `project_id` |
|
||||
| Following up on a freshly-written note | Use the returned permalink directly — it already encodes the routing |
|
||||
|
||||
## Note structure
|
||||
|
||||
BM treats `- [category]` lines as **observations** and WikiLink lines under `## Relations` as **relations**. Categories (`[decision]`, `[insight]`, `[risk]`, `[fact]`, `[todo]`, …) and relation types (`relates_to`, `implements`, `depends_on`, `blocks`, …) are open-ended — use what fits the content. YAML frontmatter is supported with `title`, `type`, `tags`, and `permalink` as standard fields; any custom fields are allowed. See the [knowledge format docs](https://docs.basicmemory.com/raw/concepts/knowledge-format.md) for the full convention.
|
||||
|
||||
```markdown
|
||||
# Clear Title
|
||||
|
||||
## Context
|
||||
Background and current situation.
|
||||
|
||||
## Key Points
|
||||
- Main insights
|
||||
- Important details
|
||||
|
||||
## Observations
|
||||
- [decision] We chose PostgreSQL for ACID guarantees
|
||||
- [insight] Users prefer social login
|
||||
- [risk] Deployment lacks rollback path
|
||||
|
||||
## Relations
|
||||
- relates_to [[Other Note Title]]
|
||||
- depends_on [[Database Choice]]
|
||||
|
||||
## Next Steps
|
||||
- [ ] Implement
|
||||
- [ ] Document
|
||||
```
|
||||
|
||||
## Behavior guidelines
|
||||
|
||||
1. **Search before answering.** If the user asks "what did we decide about X?", run `bm_search` first.
|
||||
2. **Offer to capture.** When the user shares decisions or meeting outcomes, ask: "Should I save this as a note?"
|
||||
3. **Suggest connections.** When a search returns related notes, surface them so the user knows what already exists.
|
||||
4. **Don't over-capture.** Auto-capture is already running per turn. Don't create a `bm_write` for every response — only for substantive content the user wants preserved.
|
||||
5. **Sensitive info.** Don't capture credentials or personal data without confirmation.
|
||||
|
||||
## Footgun
|
||||
|
||||
If a note's body contains literal `<memory-context>...</memory-context>` tags, Hermes's streaming output scrubber will eat those tags (and the text between paired ones) when you echo the note verbatim back to the user. Tool *inputs* are unaffected. If you must include such content, fence it in a code block.
|
||||
|
||||
## Further reading
|
||||
|
||||
Official docs live at [docs.basicmemory.com](https://docs.basicmemory.com). Every page has an AI-friendly raw markdown view at `/raw/<path>.md` (or send `Accept: text/markdown` to the canonical URL). `WebFetch` any of these when you need detail beyond what this skill covers:
|
||||
|
||||
- **[Knowledge format](https://docs.basicmemory.com/raw/concepts/knowledge-format.md)** — observation categories, relation types, frontmatter conventions.
|
||||
- **[Observations & relations](https://docs.basicmemory.com/raw/concepts/observations-and-relations.md)** — how notes form a graph that's searchable and traversable.
|
||||
- **[Memory URLs](https://docs.basicmemory.com/raw/concepts/memory-urls.md)** — title-based addressing, wildcards (`memory://docs/*`), and routing resolution order.
|
||||
- **[Projects & folders](https://docs.basicmemory.com/raw/concepts/projects-and-folders.md)** — multi-project layout, folder organization, cloud routing behavior.
|
||||
- **[Semantic search](https://docs.basicmemory.com/raw/concepts/semantic-search.md)** — how `bm_search` resolves queries (semantic + full-text).
|
||||
- **[MCP tools reference](https://docs.basicmemory.com/raw/reference/mcp-tools-reference.md)** — Basic Memory's full MCP surface (the `bm_*` tools here are a curated subset).
|
||||
- **[Cloud routing](https://docs.basicmemory.com/raw/cloud/routing.md)** — local vs cloud project modes, per-project routing setup.
|
||||
- **[llms.txt index](https://docs.basicmemory.com/llms.txt)** — full sitemap of raw markdown pages, useful when you need to look up a page not listed above.
|
||||
@@ -1,167 +0,0 @@
|
||||
"""
|
||||
Pytest configuration: stub Hermes-internal imports so the plugin loads
|
||||
without a Hermes install, and expose the loaded plugin module as a fixture.
|
||||
"""
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
import importlib.util
|
||||
import json
|
||||
import os
|
||||
import sys
|
||||
import types
|
||||
|
||||
import pytest
|
||||
|
||||
|
||||
def _stub_hermes_modules() -> None:
|
||||
"""
|
||||
The plugin imports `agent.memory_provider.MemoryProvider` and
|
||||
`tools.registry.tool_error`. These are provided by Hermes at runtime,
|
||||
not as a pip-installable package. Stub them before module load so unit
|
||||
tests don't need Hermes installed.
|
||||
"""
|
||||
agent_mod = types.ModuleType("agent")
|
||||
agent_mp_mod = types.ModuleType("agent.memory_provider")
|
||||
|
||||
class _StubMemoryProvider:
|
||||
"""Stand-in for `agent.memory_provider.MemoryProvider`."""
|
||||
|
||||
agent_mp_mod.MemoryProvider = _StubMemoryProvider
|
||||
sys.modules.setdefault("agent", agent_mod)
|
||||
sys.modules.setdefault("agent.memory_provider", agent_mp_mod)
|
||||
|
||||
tools_mod = types.ModuleType("tools")
|
||||
tools_registry_mod = types.ModuleType("tools.registry")
|
||||
|
||||
def _tool_error(msg: str) -> str:
|
||||
return json.dumps({"error": str(msg)})
|
||||
|
||||
tools_registry_mod.tool_error = _tool_error
|
||||
sys.modules.setdefault("tools", tools_mod)
|
||||
sys.modules.setdefault("tools.registry", tools_registry_mod)
|
||||
|
||||
|
||||
_stub_hermes_modules()
|
||||
|
||||
|
||||
_PLUGIN_PATH = os.path.join(os.path.dirname(os.path.dirname(__file__)), "__init__.py")
|
||||
_spec = importlib.util.spec_from_file_location("hermes_basic_memory_plugin", _PLUGIN_PATH)
|
||||
assert _spec is not None and _spec.loader is not None
|
||||
_plugin = importlib.util.module_from_spec(_spec)
|
||||
_spec.loader.exec_module(_plugin)
|
||||
|
||||
|
||||
@pytest.fixture
|
||||
def bm():
|
||||
"""The loaded plugin module."""
|
||||
return _plugin
|
||||
|
||||
|
||||
# ---- Synthetic MCP CallToolResult objects (no MCP SDK needed at test time) ----
|
||||
|
||||
|
||||
class FakeContent:
|
||||
"""Stand-in for `mcp.types.TextContent`."""
|
||||
|
||||
def __init__(self, text: str):
|
||||
self.text = text
|
||||
|
||||
|
||||
class FakeCallToolResult:
|
||||
"""Stand-in for `mcp.types.CallToolResult`."""
|
||||
|
||||
def __init__(self, content_texts, is_error: bool = False):
|
||||
self.content = [FakeContent(t) for t in content_texts]
|
||||
self.isError = is_error
|
||||
|
||||
|
||||
@pytest.fixture
|
||||
def fake_result():
|
||||
return FakeCallToolResult
|
||||
|
||||
|
||||
# ---- Actor test helpers ----
|
||||
|
||||
|
||||
class FakeSession:
|
||||
"""
|
||||
Stand-in for `mcp.ClientSession`. `call_tool` is an async coroutine
|
||||
that records every call and returns a configurable CallToolResult.
|
||||
|
||||
- default_response: dict serialized to JSON in the result text
|
||||
- hang: when True, call_tool sleeps long enough to force a client-side timeout
|
||||
- is_error: when True, return value has isError=True
|
||||
"""
|
||||
|
||||
def __init__(self, default_response=None, hang: bool = False, is_error: bool = False):
|
||||
self.default_response = default_response if default_response is not None else {"ok": True}
|
||||
self.hang = hang
|
||||
self.is_error = is_error
|
||||
self.calls: list = []
|
||||
self.was_cancelled = False
|
||||
self._stub_handlers: dict = {}
|
||||
|
||||
def stub(self, tool_name: str, handler):
|
||||
"""Register a per-tool handler. Handler receives (args) and returns a response dict (or raises)."""
|
||||
self._stub_handlers[tool_name] = handler
|
||||
|
||||
async def call_tool(self, name: str, args: dict):
|
||||
self.calls.append((name, dict(args)))
|
||||
if self.hang:
|
||||
try:
|
||||
import asyncio as _a
|
||||
|
||||
await _a.sleep(60)
|
||||
except BaseException as e:
|
||||
# Track cooperative cancellation so tests can verify cleanup
|
||||
if type(e).__name__ in ("CancelledError",):
|
||||
self.was_cancelled = True
|
||||
raise
|
||||
if name in self._stub_handlers:
|
||||
response = self._stub_handlers[name](args)
|
||||
else:
|
||||
response = self.default_response
|
||||
return FakeCallToolResult([json.dumps(response)], is_error=self.is_error)
|
||||
|
||||
|
||||
def make_scripted_actor(
|
||||
bm, session=None, raise_at_init: BaseException | None = None, fake_tools: list | None = None
|
||||
):
|
||||
"""
|
||||
Build an actor whose `_main()` is replaced with a deterministic version.
|
||||
The fake `_main` mirrors the production happy path (sets _session, _stop_future,
|
||||
fills _tools_cache, signals ready, awaits stop) but skips the stdio subprocess.
|
||||
|
||||
raise_at_init: if set, raised inside _main so we can test failure paths.
|
||||
"""
|
||||
import asyncio
|
||||
|
||||
actor = bm._BmMcpActor(["fake-bm", "mcp"])
|
||||
sess = session or FakeSession()
|
||||
|
||||
async def _fake_main():
|
||||
if raise_at_init is not None:
|
||||
actor._init_error = raise_at_init
|
||||
actor._ready.set()
|
||||
raise raise_at_init
|
||||
actor._session = sess
|
||||
actor._stop_future = asyncio.get_running_loop().create_future()
|
||||
actor._tools_cache = fake_tools or [
|
||||
{"name": n, "description": ""}
|
||||
for n in [
|
||||
"search_notes",
|
||||
"read_note",
|
||||
"write_note",
|
||||
"edit_note",
|
||||
"build_context",
|
||||
"delete_note",
|
||||
"move_note",
|
||||
]
|
||||
]
|
||||
actor._ready.set()
|
||||
await actor._stop_future
|
||||
|
||||
actor._main = _fake_main # type: ignore[assignment]
|
||||
actor._test_session = sess # convenience handle for assertions
|
||||
return actor
|
||||
Some files were not shown because too many files have changed in this diff Show More
Reference in New Issue
Block a user