session-retrospective

name: session-retrospective description: "Claude Codeのセッションを分析し、設定改善案（permissions/memory/rules/skills/agents/hooks）を提案する。完了セッションの振り返り時に使用する。トリガー例: retrospective / session analysis / improve Claude config / セッションID指定。"

Session Retrospective

Claude Codeのセッションログを分析し、設定改善案を提案する。

入力

• Session ID: 引数で渡される（例: /session-retrospective abc-123-def）
• Session File: ~/.claude/projects/*/<SESSION_ID>.jsonl
• Debug Log: ~/.claude/debug/<SESSION_ID>.txt（任意）

セッションログ形式

JSONL with records like:

{"type": "user", "message": {"role": "user", "content": "..."}}
{"type": "assistant", "message": {"role": "assistant", "content": [{"type": "tool_use", "name": "...", "input": {...}}]}}

分析手順

1. セッションファイルを読む

# セッションファイルを探す
find ~/.claude/projects -name "<SESSION_ID>.jsonl" 2>/dev/null

# セッションログを読む
cat <SESSION_FILE>

# デバッグログがあれば読む
cat ~/.claude/debug/<SESSION_ID>.txt 2>/dev/null

2. 分析して改善案を出す

Permissions（settings.json）

• Look for Permission suggestions for in debug log
• Identify commands that required manual approval
• Suggest permissions.allow rules

Memory / Rules（CLAUDE.md, .claude/rules/）

• Find repeated explanations from user
• Find AskUserQuestion tool calls with similar questions
• Suggest content for CLAUDE.md or path-specific rules

Skills（.claude/skills/<name>/SKILL.md）

• Identify repeated multi-step workflows
• Find manual instructions given multiple times
• Suggest skill definitions with frontmatter

Agents（.claude/agents/<name>.md）

• Find repeated Task tool delegations with same patterns
• Identify common subagent_type + description combinations
• Suggest custom agent definitions

Hooks（settings.json）

• Find repeated pre/post processing steps
• Identify manual formatting or validation tasks
• Suggest hook configurations

3. レポートを作成する

Save to ~/.claude/retrospectives/<SESSION_ID>.md:

# Session Retrospective: <SESSION_ID>

**Date**: YYYY-MM-DD
**Project**: /path/to/project

## Session Summary
[What was accomplished]

## What Went Well
- [Successes]

## Suggested Improvements

### 1. Permissions
```json
{"permissions": {"allow": ["Bash(command:*)"]}}

Reason: [Why this helps]

2. Memory / Rules

# Content to add

Reason: [Why this helps]

3. Skills

---
name: skill-name
description: ...
---

Reason: [Why this helps]

4. Agents

---
name: agent-name
tools: Read, Grep
---

Reason: [Why this helps]

5. Hooks

{"hooks": {...}}

Reason: [Why this helps]

Action Items

• Specific action 1
• Specific action 2

## 参照ドキュメント

- Memory: https://code.claude.com/docs/en/memory
- Skills: https://code.claude.com/docs/en/skills
- Slash Commands: https://code.claude.com/docs/en/slash-commands
- Sub-agents: https://code.claude.com/docs/en/sub-agents
- Hooks: https://code.claude.com/docs/en/hooks-guide