* changed configuration to focus on seed-gen
* added nomminally working debug_subagent_task
* debugging process
* debugging process
* debugging process
* debugging process
* got debug subagent to work more reliably
* debugging why it cant find the seed file
* added more tests
* more debugging of debugginf functionality
* more debugging of debugger
* debugging working, scripts bad because some symbols are not found
* mostly working live debugging, still a bit pricy though
* Added hybrid mode that will do batch, then try interactive if that fails
* fixes
* fixes, updating docker interactive
* updating mi parsing for gdb
* modifications to the mi parser
* Improve coverage tracking precision (#401)
* Improve coverage tracking precision
Previously coverage tracking included all regions. This changes coverage tracking to be more precise.
Macros that result in code are represented as a single line in the function, and any executed line in the macro will make that line 'covered'. Beyound that, only CodeRegions will count. This will prevent code guarded by #if that aren't in the binary from being considered reachable for example. The idea is that seed-gen would have more accurate information when selecting functions to target.
I've done some testing and it appears as if the new implementation in general reaches more lines and covers more functions.
* Fix macro coverage leaking across files due to missing filename in key
The expansion_coverage map used (line, col) as key without filename,
causing coverage from one file to incorrectly appear in another when
macros were at the same coordinates.
* Recursively expand macros and add named types for coverage data
- Process ExpansionRegion target_regions recursively instead of counting
only the call site line
- Add coordinate index for O(1) expansion lookups
- Cache computed expansion lines to avoid recomputation across functions
- Use bulk set operations for better performance
- Prevent infinite loops with visited set for circular macro references
- Add named types for coverage data structures:
- RegionCoords, ExpansionKey, CachedExpansionLines
- Type aliases: ExpansionMap, CoordToFilenames, ExpansionLinesCache
- Add comprehensive tests for nested macros and edge cases
* fix: fixing `line 110: set: -g: invalid option` for Fish shell (#408)
Co-authored-by: kevin-valerio <kevin-valerio@users.noreply.github.com>
* working state
* working state for monolith, still needs refactor
* limit grep output (whoops) and fix building to force optimization flags
* added function lookup tool
* added better build selection logic, and avoided c ode duplication
* fix fuzzer selection to ignore debug
* added debug builds as a seperate 'sanitizor'
* fixed build system, more in line with other dependancies now
* adding new better debug targets, and logging
* final changes before testing
* improving test coverage
* improving test coverage
* feat: add extract_povs command to buttercup-util (#410)
Add new CLI subcommand to extract PoVs, stack traces, and patches from
Redis submissions into a structured directory format for easy analysis.
Features:
- Extracts crash inputs (PoV files) via kubectl cp from cluster pods
- Writes fuzzer and tracer stack traces to text files
- Exports associated patches with metadata
- Organizes output by project/task_id/vulnerability
- Supports filtering by task_id and passed_only options
- Skips empty patch trackers (placeholders that never received content)
* Use git-lfs when downloading challenges
* improving test coverage
* Delete b.txt
* final changes before run hopefully
* feat: add extract_povs command to buttercup-util (#410)
Add new CLI subcommand to extract PoVs, stack traces, and patches from
Redis submissions into a structured directory format for easy analysis.
Features:
- Extracts crash inputs (PoV files) via kubectl cp from cluster pods
- Writes fuzzer and tracer stack traces to text files
- Exports associated patches with metadata
- Organizes output by project/task_id/vulnerability
- Supports filtering by task_id and passed_only options
- Skips empty patch trackers (placeholders that never received content)
* Sanitize exception messages for git clone command (#411)
* Sanitize exception messages for git clone command
* Update test case
* Don't sanitize exception if there is no PAT
* merging with main and fixing tests which tested old build system
* fixing erronous changes from testing
* fix silly linting errors
* reformatted for linter
* fixing failing tests
* fixing failing tests
* linting
* restore scripts to remove debugging changes
---------
Co-authored-by: Henrik Brodin <90325907+hbrodin@users.noreply.github.com>
Co-authored-by: Kevin Valerio <24193167+kevin-valerio@users.noreply.github.com>
Co-authored-by: kevin-valerio <kevin-valerio@users.noreply.github.com>
Co-authored-by: Ronald Eytchison <58823072+reytchison@users.noreply.github.com>
* Test configuration for Azure deployment
* ci: re-enable docker build/push
* ci: docker push labeled PRs
* Use `make deploy` for both local and azure deployments
* doc: update AKS_DEPLOYMENT doc
* add confirmation in Makefile
* Make resource group location configurable
* Add gemini api key option during setup and in deployment environments
* Add gemini pro as a fallback model in all components
* Lint
* Set rate limits for gemini models
* Add fallback models in more places
* Fix typo
* Fix kwargs expansion
* Fix instantiation of llm with callbacks
* Fixed formatting after merge
* Fix instantiation of default models
* Lint
* Fix llm creation
* Fix
* Lint
* Lint
---------
Co-authored-by: Michael D Brown <michael.brown@trailofbits.com>
* "Claude PR Assistant workflow"
* "Claude Code Review workflow"
* Apply required customizations to Claude workflows
This commit applies the necessary customizations learned from our previous Claude workflow deployment:
## claude.yml changes:
- Add Git config environment variables for private submodule authentication
- Enable submodules in checkout with persist-credentials: false
- Set 60-minute timeout
- Enable Buttercup-specific allowed tools: make lint, deployment commands, and pytest
## claude-code-review.yml changes:
- Enable sticky comments for better PR review experience
- Filter to run only on external contributors (FIRST_TIME_CONTRIBUTOR, CONTRIBUTOR, NONE)
- Add same Git authentication and submodules support
- Set 60-minute timeout
- Enable Buttercup-specific allowed tools
These changes ensure Claude can:
1. Access private submodules
2. Run necessary build/test commands
3. Provide effective code reviews for external contributors
4. Maintain review context with sticky comments
🤖 Generated with [Claude Code](https://claude.ai/code)
Co-Authored-By: Claude <noreply@anthropic.com>
* Enhance Claude workflows and improve CI efficiency
This commit applies critical customizations to Claude workflows and improves overall CI efficiency:
## Claude Workflow Enhancements
- Enable sticky comments for better PR review UX
- Add comprehensive allowed_tools list for development commands
- Keep 60-minute timeout for complex operations
- Enable submodules support for complete repository context
- Remove unnecessary Git auth (repo is now public)
- Remove author filtering to review all PRs initially
## CI Performance Improvements
- Add intelligent path filtering to lint and test workflows
- Skip CI runs for documentation-only changes
- Always run full suite on main branch
- ~70% reduction in CI minutes for non-code changes
- Add fail-fast: false to see all failures at once
- Separate fuzzer into experimental jobs with clear labeling
- lint-fuzzer-experimental
- test-fuzzer-experimental
- Makes it obvious fuzzer is allowed to fail
## Benefits
- Clearer CI status (experimental vs required)
- Faster feedback on PRs
- Reduced GitHub Actions costs
- Better debugging with all failures visible
- Claude can effectively review and assist with development
🤖 Generated with [Claude Code](https://claude.ai/code)
Co-Authored-By: Claude <noreply@anthropic.com>
* Fix CI failures for seed-gen and improve test infrastructure
This commit fixes critical CI issues and improves test infrastructure:
## Bug Fixes
- Fix coverage module name mapping for components with hyphens (seed-gen -> seed_gen, program-model -> program_model)
- Install codequery dependencies for both program-model AND seed-gen (seed-gen imports from program_model.codequery)
- Use bash parameter substitution to handle hyphen-to-underscore conversion consistently
## Safety Improvements
- Restrict git operations in Claude workflow to safe patterns only:
- git merge --ff-only (fast-forward only, no conflicts)
- git merge --no-ff --no-edit origin/* (no interactive prompts)
- git rebase --abort (can abort but not start rebases)
## Why seed-gen was failing
1. pytest-cov was looking for module "seed-gen" but Python module is "seed_gen"
2. seed-gen tests import from program_model.codequery but codequery wasn't installed
These fixes ensure all component tests run correctly with proper coverage tracking.
🤖 Generated with [Claude Code](https://claude.ai/code)
Co-Authored-By: Claude <noreply@anthropic.com>
* Optimize CI with smart path filtering and consolidated coverage
Major CI optimizations to reduce unnecessary runs and improve efficiency:
## Smart Path Filtering with dorny/paths-filter
- Added component-specific change detection
- Only runs tests/linting for components that actually changed
- Respects dependencies (e.g., common changes trigger all dependent components)
- Workflow changes trigger full suite for safety
- Main branch always runs everything
## Explicit Matrix Configuration
- Removed fragile bash transformations (tr '-' '_')
- Each component explicitly defines its coverage_module
- Matrix includes should_run conditions based on detected changes
- Cleaner, more maintainable configuration
## Consolidated Coverage Upload
- Single coverage-upload job after all tests complete
- Downloads all artifacts and uploads once to Codecov
- Reduces API calls and avoids rate limiting
- More efficient than per-component uploads
## Test Dependencies Optimization
- Reverted pytest-html/pytest-cov from component dependencies
- Install test tools with --isolated flag at CI level
- Avoids dependency duplication across components
- Prevents version conflicts
## Benefits
- ~70% reduction in CI minutes for component-specific changes
- Only affected components run tests/linting
- Single coverage upload instead of 6+ separate uploads
- Cleaner dependency management
- Better resource utilization
## Example Impact
- Changing patcher/src/foo.py now only runs patcher tests (not all 6 components)
- Changing common/ still triggers all tests (since everything depends on it)
- Documentation changes don't trigger any component tests
🤖 Generated with [Claude Code](https://claude.ai/code)
Co-Authored-By: Claude <noreply@anthropic.com>
* Trigger CI tests for patcher and seed-gen to verify fixes
* Add test trigger to Python files to verify CI fixes for patcher and seed-gen
* ci: temporarily remove path filtering to debug test failures
- Remove path filtering from tests.yml to ensure tests run
- Remove conditional execution based on path changes
- This is temporary - will re-enable after confirming our coverage fixes work
- Need to verify that buttercup.patcher and buttercup.seed_gen modules are correctly resolved
* ci: remove risky operations and unnecessary tools from Claude workflow
- Remove risky git merge --no-ff --no-edit origin/* operation
This was too broad and could merge any remote branch automatically
- Remove Docker/Kubernetes operational tools (docker ps, kubectl, helm)
Claude doesn't need direct access to running containers or clusters
- These tools are for ops tasks, not development work
- Also includes temporary removal of path filtering to debug test failures
* fix: correct helper.py path in seed-gen test fixtures
The test fixtures were creating helper.py at the wrong location:
- Was: fuzz-tooling/infra/infra/helper.py
- Now: fuzz-tooling/projects/infra/helper.py
This matches the actual path expected by ChallengeTask, fixing 4 test failures in the seed-gen component.
* fix: install codequery dependencies for patcher tests
The patcher component imports and uses program-model's codequery functionality,
so it needs the same dependencies (cscope, ctags, cqmakedb, cqsearch) installed
during CI testing.
🤖 Generated with [Claude Code](https://claude.ai/code)
Co-Authored-By: Claude <noreply@anthropic.com>
* fix: install docker-buildx-plugin for deploy-local target (fixes#265)
The docker buildx plugin is required for the deploy-local target.
This change ensures the plugin is installed regardless of whether
Docker is already installed or not.
* refactor: remove experimental label from fuzzer tests
- Integrate fuzzer into main test matrix alongside other components
- Remove separate test-fuzzer-experimental job entirely
- Split ruff and mypy steps in lint workflow for better granularity
- Keep mypy as continue-on-error with clear documentation about why
- Add warning message when mypy fails to track technical debt
The fuzzer tests have been stable with all 50 tests passing consistently.
The 'experimental' label was a vestige from earlier development when the
component had stability issues. Type checking still has known issues due
to complex external dependencies, but this is technical debt rather than
test instability.
* cleanup: remove CI trigger comments and files
- Remove '# Trigger CI test run' comments from README files
- Delete __init__.py files that were added solely to trigger CI
- These artifacts were temporary fixes to force CI runs and are no longer needed
The CI now runs properly based on path filters and these trigger
artifacts just add noise to the codebase.
* feat: implement multi-tiered integration testing strategy
- Add test-integration job to tests.yml with selective triggers
- Daily schedule at 2 AM UTC
- Manual workflow dispatch with component selection
- PR label trigger 'integration-tests'
- Tests 4 components: common, patcher, program-model, seed-gen
- Modify integration.yml triggers to be more selective
- Remove main branch push trigger
- Add weekly schedule (Sundays at 3 AM UTC)
- Add PR label trigger 'full-integration'
- Keep workflow dispatch for manual runs
- Add CI status badges to README
- Unit Tests, Integration Tests, System Integration badges
- Document integration testing strategy in CONTRIBUTING.md
- Three test tiers with timing and resource usage
- Local testing instructions
- PR labeling guidance
This avoids running expensive tests on every main push while maintaining
regular automated testing through schedules and manual control via labels.
* fix: restore seed_gen __init__.py with module_name definition
The __init__.py file was accidentally deleted in the cleanup commit,
but it contains the __module_name__ variable needed by utils.py
* security: restrict Claude workflow permissions
- Replace wildcard script execution with explicit allowed scripts
- Remove potentially risky git operations (checkout, merge, fetch, pull)
- Keep only safe git operations (status, diff, log, add, commit, push)
- Explicitly list allowed scripts for better security control
* docs: clarify base64 encoding in integration workflow
Add comment explaining that base64 encoding of GitHub token is for
Docker registry authentication format requirements, not security
---------
Co-authored-by: Claude <noreply@anthropic.com>
* feat: Add simplified SigNoz integration using official Helm chart
- Add SigNoz as a Helm dependency with conditional deployment
- Enable SigNoz by default for minikube environments
- Auto-configure OTEL to use internal SigNoz when enabled
- Update Docker Compose to include existing SigNoz stack
- Minimal configuration following the official chart approach
Co-authored-by: Riccardo Schirone <ret2libc@users.noreply.github.com>
* Configure local signoz
* do not deploy signoz in ci
* makefile: ensure signoz is available
* feat: Simplify SigNoz integration for improved user experience
- Remove external SigNoz configuration prompts from setup-local script
- Make local SigNoz deployment the default for quickstart experience
- Move external SigNoz configuration to MANUAL_SETUP.md for advanced users
- Streamline README log access section to focus on local SigNoz UI
- Relocate kubectl commands to QUICK_REFERENCE.md as alternative method
- Improve documentation structure for better user onboarding
Addresses feedback from @michaelbrownuc to simplify the integration
before merging.
🤖 Generated with [Claude Code](https://claude.ai/code)
Co-authored-by: Riccardo Schirone <ret2libc@users.noreply.github.com>
* Update MANUAL_SETUP.md
---------
Co-authored-by: claude[bot] <209825114+claude[bot]@users.noreply.github.com>
Co-authored-by: Riccardo Schirone <ret2libc@users.noreply.github.com>
* feat: save CRS artifacts to run-data directories
Add functionality to automatically save POVs, patches, and bundles
received from the CRS into organized run-data-<timestamp> directories
structured by task-id as requested in issue #205.
Changes:
- Add run_data_dir configuration setting to Settings
- Implement save_artifact() utility function with proper base64 decoding
- Modify bundle, patch, and POV endpoints to save artifacts to disk
- Create directory structure: run-data-YYYYMMDDHHMMSS/<task-id>/{povs,patches,bundles}/
- Handle different file types: .json for bundles, .patch for patches, .bin for POVs
Co-authored-by: Riccardo Schirone <ret2libc@users.noreply.github.com>
* feat: add hostPath volume for buttercup-ui run data access
- Add run-data-volume hostPath mount to UI deployment for minikube
- Maps container /tmp/buttercup-run-data to host /data/buttercup-run-data
- Enables external access to CRS artifacts (POVs, patches, bundles)
- Falls back to emptyDir for non-minikube environments
Co-authored-by: Riccardo Schirone <ret2libc@users.noreply.github.com>
* fix: add BUTTERCUP_UI_RUN_DATA_DIR environment variable to UI deployment
- Set BUTTERCUP_UI_RUN_DATA_DIR to /tmp/buttercup-run-data to match volume mount
- Ensures UI application saves run data artifacts to the correct directory
- Run data will now be accessible outside Kubernetes cluster via hostPath volume
Co-authored-by: Riccardo Schirone <ret2libc@users.noreply.github.com>
* fix save-artifact feature
* orchestrator: fix lint
---------
Co-authored-by: claude[bot] <209825114+claude[bot]@users.noreply.github.com>
Co-authored-by: Riccardo Schirone <ret2libc@users.noreply.github.com>
Co-authored-by: Michael D Brown <michael.brown@trailofbits.com>
* Fix integration-test run
* use the proper oss-fuzz-aixcc commit to run integration-test
* determine OSS_FUZZ_CONTAINER_ORG at runtime
* use `git diff` in the patcher to create the patch to fix a problem
with patches affecting non-newline terminated files
* fix parsing
* fix program-model lint
* Modify CRS to work on MacOS/ARM
* Use LibUCL for testing
* use example-libpng
* fix linting
* common: fix tests
* Remove just tool completely and migrate all functionality to Makefiles
- Added install-cscope, lint-python, and lint-python-all targets to main Makefile
- Created program-model/Makefile with all justfile functionality
- Removed justfiles from both main and program-model directories
- Updated all documentation to use make commands instead of just
- Removed just installation and checking functions from setup scripts
- All functionality preserved with equivalent make targets
Co-authored-by: Riccardo Schirone <ret2libc@users.noreply.github.com>
* Remove duplicate lint-python targets from Makefile
These targets duplicated existing lint and lint-component targets.
Updated CLAUDE.md to reference the correct make targets.
Co-authored-by: Riccardo Schirone <ret2libc@users.noreply.github.com>
* Fix remaining references to removed lint-python targets in documentation
Co-authored-by: Riccardo Schirone <ret2libc@users.noreply.github.com>
* Fix Makefile: Add missing targets to .PHONY declaration
Added lint, lint-component, clean-local, wait-crs, check-crs,
crs-instance-id, status, and send-integration-task to .PHONY
to ensure they work correctly even if files with those names exist.
Co-authored-by: Riccardo Schirone <ret2libc@users.noreply.github.com>
* ci: remove just references
---------
Co-authored-by: claude[bot] <209825114+claude[bot]@users.noreply.github.com>
Co-authored-by: Riccardo Schirone <ret2libc@users.noreply.github.com>
Co-authored-by: Michael D Brown <michael.brown@trailofbits.com>
* add 'status' make command to check deployment status
* unify make commands
* rename `make test` to `make send-integration-task`
* add check-crs message
By default, the bash installer on linux installs to ~/bin and needs to be added to PATH manually.
I'm not sure how it behaves on other platforms, but it might be a smoother experience to rely on
the platform's default package manager to install just to the right place.
Co-authored-by: Michael D Brown <michael.brown@trailofbits.com>
* deployment: just use the value in values.template
* ci: disable integration tests and private settings
* Download trailofbits cscope, not aixcc-finals one
* ci: fix docker login to ghcr.io
* Changes to allow non aixcc deployment
* buttercup-ui: basic skeleton for a CRS interface
* file-server implementation
* add ui to k8s
* other apis
* try to fix k8s
* remove some labels
* fix ui
* small fixes to doc
* ui: support for cloning private repos
* Add setup scripts for easy deployment
* update scripts
* update make/readme
* small adj
* address review
* we need 0.0.0.0 for k8s
IsaacTrost