メニュー
Archive a skill instead of deleting it — fully reversible. Use when the user says "retire skill", "archive skill", "deactivate skill", "disable skill", or "skill'i kaldır".
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
SOC 職業分類に基づく
Analyze accumulated feedback memories and turn recurring patterns into permanent controls (rules, permissions, hooks) or common-mistakes entries. Use when the user says "improve", "iyileştir", "apply feedback", "kurallaştır", or "turn feedback into rules".
Project onboarding and template setup. Use when the user says onboard, setup, initialize, init, configure this project, or first-time setup. Guides through detecting project mode and stack, configuring CLAUDE.md, permissions, hooks, rules, and architecture docs.
Save session learnings, patterns, and corrections to persistent memory. Use when the user says "save learnings", "save memory", "remember this session", or at the end of a productive session.
Add a new Claude Code control following project conventions. Use when the user says "new rule", "add rule", "create rule", "yeni rule ekle", or wants to enforce behavior consistently. Guides through choosing enforcement surfaces, writing effective rule files, and keeping controls consistent.
Create a new Claude Code skill following project conventions. Use when the user says "new skill", "add skill", "create skill", or "yeni skill ekle". Guides through gathering requirements, writing effective SKILL.md files, and structuring skill directories.
Add a new Claude Code hook — shell script wired into settings.json. Use when the user says "new hook", "add hook", "yeni hook ekle", or wants to automate something that runs on every tool use, session start, or stop.
| name | retire-skill |
| description | Archive a skill instead of deleting it — fully reversible. Use when the user says "retire skill", "archive skill", "deactivate skill", "disable skill", or "skill'i kaldır". |
| argument-hint | <skill-name> |
| allowed-tools | Read, Glob, Edit, Bash(git mv *), Bash(mkdir *) |
| disable-model-invocation | true |
Target skill: $ARGUMENTS
If no argument given, Glob .claude/skills/ and list available skills, then ask which to retire.
Otherwise check for .claude/skills/<name>/SKILL.md. If not found, stop and say so.
Before archiving, check if any Type B agent's AGENT.md references this skill:
.claude/agents/ for directoriesAGENT.md and scan the Skills tableWarning: <skill-name> is listed in <agent-name>'s AGENT.md Skills table.
Retiring it will leave a broken reference.
Options:
a) Remove from AGENT.md too (recommended)
b) Keep the reference and archive anyway
c) Cancel
Wait for answer. If (a), edit the parent AGENT.md to remove the row.
Retiring: <skill-name>
This will move the skill to .claude/skills/archive/<skill-name>/.
It will no longer appear in skill lists or be triggered.
Nothing is deleted — move it back to .claude/skills/ to restore.
Proceed? (yes / no)
Wait for confirmation.
Ensure .claude/skills/archive/ exists (create if needed).
git mv .claude/skills/<name> .claude/skills/archive/<name>
If git mv is unavailable (untracked files), use Bash mv instead.
Retired: <skill-name>
Archived to: .claude/skills/archive/<skill-name>/
[Removed from: .claude/agents/<agent-name>/AGENT.md Skills table] ← if applicable
To restore: move back to .claude/skills/