learning-log

Document debugging discoveries by capturing category, title, description, and affected files into a structured learning log at .claude/learning-log.md. This records non-obvious insights like bug root causes and framework surprises for later search and review.

Track architecture decisions and performance investigations by writing date-stamped entries with descriptive fields into a structured learning log. This preserves workaround rationale and surprising behaviors so teams avoid repeating mistakes and improve future decisions.

Keep learning insights separate by storing descriptive discoveries in .claude/learning-log.md, while keeping prescriptive rules in MEMORY.md and temporary handoff state in .claude/handoff.md. This differentiation ensures descriptive findings remain distinct from actionable rules and session context.

Yes, a learning log records external service discrepancies and framework surprises using structured categories like External Service and Gotcha. Each entry captures what happened, what was tried, and what worked, ensuring non-obvious findings are searchable for future reference.

Structure development knowledge management entries using categories like Architecture Decision, Bug Root Cause, Performance Discovery, Pattern Found, Gotcha, and External Service. These categories improve retrieval by organizing non-obvious findings into distinct, searchable classifications.

Teams repeat debugging mistakes when non-obvious insights and workaround rationale are lost. Capturing findings into a structured learning log with consistent fields prevents this by creating a searchable record of bug root causes and surprising behaviors for later review.