Files
David Maynor 11c23bb4a6 fix: use >- folded scalar for SKILL.md descriptions (#20)
The literal block scalar (description: |) preserves newlines, causing
multi-line descriptions to render poorly in the Claude Code /skills
menu. This converts all 5 affected SKILL.md files to use the folded
strip scalar (>-), which collapses to a single line at parse time.

Also updates the skill-template reference to recommend >- so future
skills don't reintroduce the issue.

Affected plugins:
- skill-extractor
- ffuf-web-fuzzing
- humanizer
- planning-with-files
- wooyun-legacy
2026-02-22 22:09:49 -08:00
..

skill-extractor

Extract reusable skills from work sessions. Manual invocation only - no hooks, no noise.

Usage

/skill-extractor [--project] [context hint]

Invoke after solving a non-obvious problem to capture the knowledge as a reusable skill.

Examples:

/skill-extractor                           # Extract from current session
/skill-extractor --project                 # Save to project instead of user level
/skill-extractor the cyclic data DoS fix  # Hint to focus extraction

What It Does

  1. Analyzes your conversation for extractable knowledge
  2. Presents a quality checklist for confirmation
  3. Optionally researches best practices (web search, Context7)
  4. Generates a skill following repository standards
  5. Validates structure before saving
  6. Saves to ~/.claude/skills/ or .claude/skills/

Quality Gates

Before extraction, you confirm:

  • Reusable (helps future tasks)
  • Non-trivial (required discovery)
  • Verified (solution worked)
  • Specific triggers (exact errors/scenarios)
  • Explains WHY (not just steps)

Installation

/plugin install trailofbits/skills-curated/plugins/skill-extractor