From kimchi
Runs Kimchi self-improvement: observes bead executions or feedback, proposes minimal updates to skills/validators/personas/templates, validates against history, and applies with git versioning.
How this skill is triggered — by the user, by Claude, or both
Slash command
/kimchi:halmoniThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
> "Taste, adjust, remember. Each batch teaches the next."
"Taste, adjust, remember. Each batch teaches the next."
<command_purpose> Self-improvement system that observes what happened during bead execution or from user feedback, proposes small improvements to skills/validators/personas, validates them, and applies with versioning. </command_purpose>
Parse $ARGUMENTS:
--from-bead ID: Analyze execution of a specific bead--from-feedback "text": Process explicit user feedback--taste-only: Observe and report without proposing changes--dry-run: Show proposals without applying--history: Show skills/CHANGELOG.mdFrom bead execution (--from-bead):
.beads/{id}.yaml for the bead specFrom feedback (--from-feedback):
Interactive (no args):
Report observations:
Tasting...
Observations:
- [What happened]
- [What worked well]
- [What could be better]
Generate specific, minimal changes:
Proposed adjustment:
File: [path to skill/validator/persona]
Change:
+ [line to add]
- [line to remove]
Rationale: [Why this helps, with evidence]
Rules:
Check the proposal:
| Check | Question |
|---|---|
| Conflicts | Does this contradict existing skills/validators? |
| Specificity | Is this actionable, not vague? |
| Generality | Will this help beyond this one case? |
| Protected | Does this touch protected components? |
Protected components (CANNOT modify):
If proposal touches protected components, reject it and explain why.
If --dry-run: Show proposal and stop.
If --taste-only: Show observations and stop.
Otherwise, ask for confirmation:
Apply this change? [y/n/edit]
If confirmed:
skills/CHANGELOG.mdhalmoni: [brief description]--history)If --history flag, just read and display skills/CHANGELOG.md.
npx claudepluginhub tromml/kimchi --plugin kimchiImproves Claude Code skills post-use: diagnoses issues from execution (outdated/missing/unclear/wrong), previews diffs, edits SKILL.md after confirmation, logs changes.
Reviews session to propose lean updates to existing skills, capturing debugging patterns, data issues, queries, doc references, and workflow improvements. Invoke at session end.
Retrospectively evaluates and improves skills, agents, commands, and hooks using skill logs, git history, knowledge base, and performance metrics. Use after slow, confusing, or fragile sessions.