← ClaudeAtlas

self-reflectlisted

End-of-session reflection that extracts learnings, patterns, gotchas, and decisions into the knowledge base. Run after completing significant work to make future sessions smarter.
antonioshaman/aura-companion · ★ 0 · AI & Automation · score 76
Install: claude install-skill antonioshaman/aura-companion
Analyze the work done in this session and extract valuable learnings into the knowledge base. ## Process ### 1. Gather Context Review what happened in this session: - **Read recent git log** to see what was committed - **Check git diff** for uncommitted changes - **Review any test failures** encountered during the session - **Note any corrections** the user made to your approach ### 2. Identify Learnings Extract insights across these categories: | Category | File | What to capture | |----------|------|----------------| | **Patterns** | `.agents/knowledge/patterns.jsonl` | Reusable approaches that worked well | | **Gotchas** | `.agents/knowledge/gotchas.jsonl` | Surprising behaviors, tricky edge cases | | **Decisions** | `.agents/knowledge/decisions.jsonl` | Architectural choices and their rationale | | **Anti-patterns** | `.agents/knowledge/anti-patterns.jsonl` | Approaches that failed or caused issues | | **Codebase facts** | `.agents/knowledge/codebase-facts.jsonl` | New structural knowledge about the repo | | **API behaviors** | `.agents/knowledge/api-behaviors.jsonl` | Model/tool/API quirks discovered | ### 3. Write Entries For each learning, append a JSONL entry. Field order matches `.agents/knowledge/README.md`: ```json { "id": "<category-prefix>-<next-number>", "type": "<pattern|gotcha|decision|anti-pattern|codebase-fact|api-behavior>", "fact": "Core insight in one sentence", "recommendation": "How to apply this in future sessions", "confidence": "<