| name | learn |
| description | Analyze conversation for learnings and save to docs folder |
| argument-hint | optional topic hint |
| model | claude-opus-4-5-20251101 |
| allowed-tools | Read, Write, Edit, Glob, Grep, AskUserQuestion |
Learn from Conversation
Analyze this conversation for insights worth preserving in the project's documentation.
If a topic hint was provided via $ARGUMENTS, focus on capturing that specific learning.
If no hint provided, analyze the full conversation for valuable insights.
Phase 1: Deep Analysis
Think deeply about what was learned in this conversation:
- What new patterns or approaches were discovered?
- What gotchas or pitfalls were encountered?
- What architecture decisions were made and why?
- What conventions were established?
- What troubleshooting solutions were found?
Only capture insights that are:
- Reusable - Will help in future similar situations
- Non-obvious - Not already common knowledge
- Project-specific - Relevant to this codebase
If nothing valuable was learned, say so and exit gracefully.
Phase 2: Categorize & Locate
Read existing docs to find the best home. Look for a docs/ folder or similar documentation directory in the project.
If no existing doc fits, propose a new doc file with kebab-case naming.
Note: CLAUDE.md stays stable as the entry point. All detailed learnings go to /docs only.
Phase 3: Draft the Learning
Format the insight to match existing doc style:
- Clear heading describing the topic
- Concise explanation of the insight
- Code examples if applicable
- Context on when this applies
Phase 4: User Approval (BLOCKING)
Present your proposed changes:
- What insight you identified
- Where you'll save it (existing doc + section, or new file)
- The exact content to add
Wait for explicit user approval before saving.
Phase 5: Save
After approval, save the learning and confirm what was captured.