forked from bchanot/claude
6,416 → 4,243 chars (≈ −540 tokens/session, catalog loaded every session). Kept per BDR-014/LRN-043: 'Use when' pattern, discriminating FR+EN triggers, all 'For X → /Y' disambiguation lines. Cut: redundant trigger synonyms, header/engine enumerations, prose the model derives. find-docs excluded (ctx7-owned, regen clobbers — LRN-086); doc + geo taken instead. All ≤ ~505 chars body (BDR-014 aspirational ceiling). Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_016zA3Qh2Q1QpcGXzXxKeDHR
900 B
900 B
name, version, description, allowed-tools
| name | version | description | allowed-tools | ||||||
|---|---|---|---|---|---|---|---|---|---|
| commit-change | 1.0.0 | Analyze all pending changes (staged, unstaged, untracked) and create atomic commits grouped by logical unit, retracing the work. Any git repository. Triggers: "commit my changes", "smart commit", "auto commit", "commit everything", "analyse et commit", or any variation of committing messy pending work intelligently. |
|
Load and follow strictly: $HOME/.claude/agents/commit-changer.md.
If unreachable, emit Commit-changer agent missing. and STOP. Never auto-commit blind — a wrong group is harder to undo than not committing.
Pre-flight checks (the agent should also perform, but flag here):
- Detached HEAD or unmerged conflicts → STOP, report state.
- Identity unconfigured (
git config user.emailempty) → STOP, ask user.
$ARGUMENTS