{
  "permissions": {
    "deny": [
      // Secrets: 物理的に隔離するのが第一。念のため全面拒否
      "Read(.env*)", "Edit(.env*)", "Write(.env*)",
      "Read(**/.env*)", "Edit(**/.env*)", "Write(**/.env*)",

      "Read(secrets/**)", "Edit(secrets/**)", "Write(secrets/**)",
      "Read(**/secrets/**)", "Edit(**/secrets/**)", "Write(**/secrets/**)",

      // よくある資格情報の場所と拡張子
      "Read(**/.ssh/**)", "Edit(**/.ssh/**)", "Write(**/.ssh/**)",
      "Read(**/*id_rsa*)", "Read(**/*.pem)", "Read(**/*.key)", "Read(**/*.p12)",
      "Read(**/*service-account*.json)", "Read(**/credentials/*.json)",

      // 外部送信経路を縮める（必要に応じて調整）
      "WebFetch",
      "Bash(curl:*)", "Bash(wget:*)", "Bash(scp:*)", "Bash(ssh:*)",
      "Bash(git push:*)"

       // その他危険なコマンド
      "Bash(git config *)",
      "Bash(brew install *)",
      "Bash(chmod 777 *)",
      "Bash(rm -rf /*)",
      "Bash(gh repo delete:*)"
    ],
    "allow": []
  }

/permissions

> .env.testというファイルを読み込んでください

⏺ Read(.env.test)
  ⎿  Error: Permission to read /Users/hoge/dev/.env.test has been denied.

⏺ .env.testファイルの読み込み権限がありません。

⏺ Bash(git config --list | grep user)
  ⎿  Error: Bash operation blocked by hook:
     - [python ~/.claude/scripts/deny-check.py]: Error: コマンドが拒否されました: 'git
      config --list | grep user' (パターン: 'Bash(git config *)')

# プロジェクト配下に作る（チーム共有に向く）
mkdir -p .claude/commands

# 個人常用コマンドはホーム配下
mkdir -p ~/.claude/commands

---
description: Analyze this code for performance issues and suggest optimizations
---

Analyze this code for performance issues and suggest practical optimizations.

---
description: Fix an issue with the provided number following our coding standards
argument-hint: [issueNumber]
---

Fix issue #$ARGUMENTS following our coding standards.
- Write tests first (TDD).
- Keep changes minimal and focused.
- Produce a short commit message and a longer body with rationale.

.claude/commands/frontend/component.md   # => /frontend:component (project)
~/.claude/commands/component.md          # => /component (user)

---
allowed-tools: Bash(git status:*), Bash(git diff:*), Bash(git branch:*), Bash(git log:*)
description: Create a conventional commit from current changes
argument-hint: [commitTitle]
---

## Context (auto-collected)
- Current branch: !`git branch --show-current`
- Git status: !`git status --porcelain`
- Diff (staged & unstaged): !`git diff HEAD`
- Recent commits: !`git log --oneline -10`

## Task
Using the context, craft a Conventional Commit:
- Title: "$ARGUMENTS"
- Body: why, what changed, risk, rollback plan
- Include a bullet list of changed files with short rationale.

---
description: Security review for the provided files and configs
---

Review the implementation for security vulnerabilities.

Focus:
- Auth flow, secrets handling, SSRF, RCE, SQL/NoSQLi, XSS, CSRF, authZ.

Reference files:
- @src/server/auth.ts
- @src/pages/api/*
- @next.config.js
- @package.json

Produce:
- Findings (severity with rationale)
- Concrete fixes (code/block diffs)
- Quick checks that can be automated in CI

---
allowed-tools: Bash(npm run:*), Bash(pnpm:*), Bash(yarn:*)
description: Run lint, summarize, and propose auto-fixes
---

## Context
- Package scripts: @package.json
- ESLint config: @.eslintrc.*, @eslint.config.js

## Run
!`npm run -s lint || true`

## Task
1) Summarize errors/warnings by rule and file (top offenders).
2) Suggest config-level fixes (but avoid masking real issues).
3) Offer safe code changes (diff blocks) for the top 5 issues.

---
allowed-tools: Bash(npm run:*), Bash(pnpm:*), Bash(yarn:*), Bash(git diff:*), Bash(git log:*)
description: Run tests, summarize failures, and propose fixes
---

## Context
- Changed files: !`git diff --name-only HEAD~1`
- Recent commits: !`git log --oneline -5`
- Vitest config: @vitest.config.*

## Run
!`npm run -s test -- --reporter=verbose || true`

## Task
- Group failures by suite/test.
- Hypothesize root causes with code pointers.
- Provide minimal patch diffs to make tests pass.

---
allowed-tools: Bash(git branch:*), Bash(git diff:*), Bash(git rev-parse:*), Bash(git log:*)
description: Generate a PR description using conventional sections
argument-hint: [title]
---

## Context
- Branch: !`git branch --show-current`
- Range: !`git rev-parse --short HEAD~10` .. !`git rev-parse --short HEAD`
- Diff summary: !`git diff --stat HEAD~1`

## Task
Create a PR body:

### Title
$ARGUMENTS

### Summary
- what/why/background

### Changes
- bullet list with scope per file

### Risks & Rollback
- risk matrix + revert steps

### Tests
- coverage and key cases

---
description: Generate release notes from tags and keywords
argument-hint: [version | tagsCSV | summary]
---

Parse $ARGUMENTS into:
- version (before first '|')
- tagsCSV (between first and second '|')
- summary (after second '|')

If a part is missing, infer from recent commits: !`git log --oneline -50`
Output a well-structured CHANGELOG section for version.

---
model: sonnet
description: Deep architectural review
---

---
description: Check code against our Clean Architecture/DDD rules
---

Review the repository for rule violations.

Rules:
- Domain entities must not import from application or infrastructure.
- Adapters may depend inward, never outward.
- UI components must not access repositories directly.
- Value Objects are immutable; no public setters.

Reference:
- @src/domains/**
- @src/app/**
- @src/infrastructure/**
- @src/components/**

Output:
- Violations list (file,line,rule)
- Suggested refactors and file moves

---
description: Draft GA4 event plan with parameters, GTM mapping, and validation steps
argument-hint: [featureName]
---

Create a GA4 event plan for "$ARGUMENTS".

Deliver:
- Event names, parameters, and types
- GTM dataLayer schema and triggers
- DebugView / Tag Assistant validation checklist
- BigQuery export table expectations

---
allowed-tools: Bash(dbt:*), Bash(gcloud:*), Bash(bq:*), Bash(gsutil:*)
description: Prepare a safe run plan for dbt models
argument-hint: [modelOrTag]
---

## Context
- dbt project: @dbt_project.yml
- Target profiles: @profiles.yml
- Recent runs (if available): !`dbt ls --resource-type model --select $ARGUMENTS || true`

## Task
Propose a safe run plan:
- Impacted models (graph)
- Test strategy (generic + singular)
- Backfill range (if incremental)
- Rollback plan