| name | reflect |
| description | Post-implementation retrospective. Run after completing any non-trivial task to extract patterns, surprises, and lessons and persist them to rules and memory without waiting for user feedback. Use proactively after creating a PR or finishing a significant feature. |
Reflect
After completing a significant implementation, ask these questions about the work just done and update the knowledge base with anything that should fire proactively next time.
Process
- Review what was just built: run
git diff main...HEAD to survey the full change
- Ask the five questions (answer each briefly before deciding whether to save):
- Pattern: Did I invent or repeat a pattern that should be a named rule so I don't have to re-derive it?
- Surprise: Did anything behave unexpectedly — a crate API, a Ratatui layout quirk, a storage edge case? Would a note have saved time?
- Struggle: Did I spend more than one round-trip on something that should have been obvious? What would make it obvious next time?
- Missed opportunity: Is there something I didn't do that I would have done if I'd noticed it earlier (refactoring, test coverage, doc gap)?
- Stale knowledge: Did anything in memory/rules turn out to be wrong or outdated?
- Classify and save each finding using the same destinations as
learn-from-feedback:
- Coding pattern →
.claude/skills/coding-standards/SKILL.md
- Domain fact →
.claude/rules/domain.md
- Testing practice →
.claude/rules/testing.md
- Workflow change →
CLAUDE.md or relevant skill
- Project-specific detail →
memory/MEMORY.md
- Stale entry → edit or remove the outdated line in place
- Summarize: Report what was saved and where (or "Nothing new to record" if nothing qualified)
Rules
- Only save findings that would have changed behavior on this task — not observations that are already obvious
- Be conservative: one clear bullet per finding, no speculative generalizations
- Don't duplicate existing entries — check before writing
- "Nothing new to record" is a valid and common outcome; do not force findings