Update an existing handoff document with current session progress. Applies section-specific merge rules: append-only for Work Done (never deletes history), replace for Status and Next Steps, merge for Files and Discoveries. Falls back to creating a new handoff if no source file is found.
インストールする前に、エージェントに与えられる指示の中身を確認できます。
Update an existing handoff document with this session's progress.
Determine which file to update, in this priority order:
$ARGUMENTS[0] was provided, use it as the file path./handoff:resume, look in conversation history for the argument that was passed to it.claudedocs/handoffs/handoff_YYYYMMDD_HHMMSS.md. Inform the user that no existing handoff was found.Read the resolved file completely. Parse all sections as the baseline for the merge.
Rewrite the file with updated content using these merge rules:
| Section | Rule | Notes |
|---|---|---|
| Task | Keep original | Update only if scope fundamentally changed |
| Scope | Keep or refine | Narrow it if new understanding emerged |
| Files | Merge | Add new files touched in this session. Keep originals. Use path:line format |
| Discoveries | Append | Add new findings below existing list. Never remove prior discoveries |
| Work Done | Append only | Add new completed items below existing list. Never remove entries. Include commit hashes |
| Status | Replace | Write the current state: what is done now, what remains, test status |
| Next Steps | Replace | Write the updated actionable checklist |
| Code | Update | Replace with the most relevant snippets from the current state |
The append-only rule for Work Done is strict. Even if a previous entry describes work that was later revised, keep it. The history is the record. Add a new entry describing the revision instead of removing the old one.
After saving, confirm:
This command implements the Handoff Triad pattern documented in this guide's Session Handoff Pattern section. Template inspired by Packmind's handoff commands (Apache 2.0).
まだレビューはありません。使ってみた感想をお寄せください。
概要と使いどころ
Audit Claude Code agents, skills, and commands for quality and production readiness. Use when evaluating skill quality, checking production readiness scores, or comparing agents against best-practice templates.
日本語の概要は準備中です。原文の説明を表示しています。
Codebase health audit scoring 7 categories with progression plan
日本語の概要は準備中です。原文の説明を表示しています。
Autonomous improvement loop: scan codebase metrics, scaffold experiment files, run agent-driven iterations until metric improves
日本語の概要は準備中です。原文の説明を表示しています。
Generate bounded independent candidates, score them against a frozen rubric, and verify the selected result with a proof log.
日本語の概要は準備中です。原文の説明を表示しています。
Post-deploy monitoring: watch production after a deploy and alert on regressions
日本語の概要は準備中です。原文の説明を表示しています。
Restore context after /clear by summarizing recent work and project state
日本語の概要は準備中です。原文の説明を表示しています。