Claude AI Configuration

A comprehensive configuration system for Claude AI with custom commands, development principles, and workflow optimizations.

📌 日本語版

🎯 Overview

This repository contains personal configurations for Claude AI, including:

• Custom slash commands for systematic development workflows (47 skills)
• Specialized AI agents for code review, generation, and analysis (38 agents)
• Core AI operation principles and development best practices
• Quality pipeline hooks (guardrails, formatter, reviews, gates)
• Japanese language support

📁 Structure

.claude/
├── CLAUDE.md              # Main configuration (AI reads this)
├── README.md              # This file - Quick start guide
├── adr/                   # Architecture Decision Records
├── rules/                 # Rule definitions
│   ├── core/             # Core AI operation principles
│   ├── conventions/      # Documentation conventions
│   ├── development/      # Development patterns & methodologies
│   ├── frameworks/       # Framework-specific rules
│   └── workflows/        # Workflow guides
├── skills/               # Skill-based knowledge modules (47 skills)
├── agents/               # Specialized AI agents (33 agents)
│   ├── architects/       # Feature architecture designers
│   ├── critics/          # Finding challengers (devils-advocate)
│   ├── enhancers/        # Code enhancers & simplifiers
│   ├── evaluators/       # Quality evaluators
│   ├── explorers/        # Codebase exploration agents
│   ├── generators/       # Code/test/git generators
│   ├── resolvers/        # Build error resolvers
│   ├── reviewers/        # Code review agents (20 reviewers)
│   └── teams/            # Integrators & implementers
├── docs/                  # Design docs & guides
├── hooks/                 # Pre/Post tool-use hooks
├── scripts/               # Utility scripts
├── output-styles/         # Output style definitions
├── .claude-plugin/        # Plugin marketplace config
└── .ja/                   # Japanese translations

🚀 Quick Start

Option 1: Install as Claude Code Plugin (Recommended)

This repository is available as a Claude Code plugin, allowing you to easily install specific workflow sets:

1. Add this repository as a marketplace:

   /plugin marketplace add thkt/dotclaude
   

2. Browse available plugins:

   /plugin
   

3. Install specific plugin (choose one or more):

   /plugin install complete-workflow-system  # Full TDD/RGRC workflow
   /plugin install quick-actions             # /fix, /polish
   /plugin install git-utilities             # /commit, /branch, /pr, /issue
   /plugin install documentation-tools       # /adr, /docs
   /plugin install browser-workflows         # agent-browser (E2E via /code)
   

Available Plugins:

• complete-workflow-system: Full development workflow with quality gates (/think, /code, /audit, /research, /feature, /swarm)
• quick-actions: Fast bug fixes (/fix), AI slop removal (/polish)
• git-utilities: Git workflow helpers (/commit, /branch, /pr, /issue, /preview)
• documentation-tools: ADR creation (/adr) and domain glossary (/glossary)
• browser-workflows: E2E testing via /code E2E Phase (agent-browser)
• productivity-tools: Inbox aggregation from GitHub, Slack, Calendar (/inbox)
• development-skills: 21 reference skills for TDD, principles, patterns, security, doc lookup, and more

Option 2: Manual Installation (Full Configuration)

For using this as your personal .claude configuration:

1. Clone this repository to your home directory:

   git clone https://github.com/thkt/.claude ~/.claude
   

2. Or if you already have a .claude directory, back it up first:

   mv ~/.claude ~/.claude.backup
   git clone https://github.com/thkt/.claude ~/.claude
   

Note: Manual installation includes all commands, agents, rules, and personal configurations. Plugin installation only includes shared commands and agents (excludes personal CLAUDE.md, rules/, and settings.json).

📦 Dependencies & Setup

Sandbox Feature (Optional but Recommended)

Claude Code's sandbox feature provides secure command execution with automatic permission handling, reducing approval fatigue while maintaining safety.

System Requirements:

• macOS or Linux (Windows not yet supported)
• Node.js with npm/npx
• ripgrep (typically pre-installed)
• jaq (for IDR hooks): brew install jaq

Setup:

# 1. Install sandbox runtime
npm install -g @anthropic-ai/sandbox-runtime

# 2. Verify installation
srt --version

# 3. Enable in Claude Code
# Run this command in Claude Code session:
/sandbox
# Select option 1: "Sandbox BashTool, with auto-allow in accept edits mode"

What it does:

• ✅ Restricts file system access to allowed directories
• ✅ Controls network access via proxy
• ✅ Auto-executes safe commands in sandbox
• ✅ Requests approval only when sandbox restrictions are hit

Configuration (optional):

Create ~/.srt-settings.json for custom settings:

{
  "sandbox": {
    "enabled": true,
    "autoAllowBashIfSandboxed": true,
    "excludedCommands": ["docker"],
    "network": {
      "allowLocalBinding": true,
      "httpProxyPort": 8080
    }
  }
}

Hook Tools (Recommended)

Quality pipeline hooks that run automatically during Claude Code sessions. These catch lint errors, format code, inject static analysis, and enforce quality gates - all without manual intervention.

brew tap thkt/tap
brew install guardrails formatter reviews gates

Tool	Hook	Timing	Role
guardrails	PreToolUse	Before Write/Edit	Lint (oxlint) + security checks
formatter	PostToolUse	After Write/Edit	Auto-format (oxfmt)
reviews	PreToolUse	Before Skill	Static analysis context injection
gates	Stop	Agent completion	Quality gates (knip, tsgo, madge)

Per-project configuration is done via .claude/tools.json. See thkt/tap for details.

External CLI Tools (Optional)

Some commands use external CLI tools for data source integration:

Tool	Required By	Purpose	Install
gh	/inbox (GitHub)	GitHub API access	brew install gh && gh auth login
gemini	/inbox (Calendar)	Google Calendar query	Gemini CLI
scout	Slack URL reading	Slack message fetch	brew install thkt/tap/scout
SLACK_TOKEN	/inbox (Slack)	Slack search API	See below

Slack reading: scout fetch <slack-url> reads any Slack message/thread URL directly. No additional setup needed if scout is configured.

Slack search (for /inbox):

1. Create a Slack App and add search:read to User Token Scopes

2. Obtain the User OAuth Token (xoxp-...)

3. Set environment variables:

   export SLACK_TOKEN="xoxp-..."
   export SLACK_WORKSPACE="your-workspace"  # the workspace part of {workspace}.slack.com
   

Required Plugins

Some commands depend on external plugins that are not included in this repository. Install them manually after cloning:

Plugin	Required By	Purpose	Install Command
ralph-loop	/code	TDD Green Phase auto-iteration	/plugin install ralph-loop

Quick Install:

/plugin install ralph-loop

Note: Plugins are stored in ~/.claude/plugins/ which is excluded from git. Each user must install plugins independently.

📝 Available Commands

See the complete command reference:

• English Command Reference
• 日本語コマンドリファレンス

🔄 Standard Workflows

Feature Development (Enhanced)

/research → /think → /code → /audit

Bug Investigation & Fix

/research → /fix

🌏 Language Support

• AI Processing: English internally
• User Output: Japanese (configurable)
• Documentation: Available in both English and Japanese

🛠️ Key Features

Core AI Principles

• Safety First: File deletion uses trash (~/.Trash/), destructive operations require confirmation
• User Authority: Your instructions are the ultimate authority
• Output Verifiability: Claims backed by evidence (file paths, confidence markers ✓/→/?)

Development Approach

• Occam's Razor: Choose the simplest solution that works
• Progressive Enhancement: Build simple, enhance gradually
• TDD/RGRC: Red-Green-Refactor-Commit cycle for reliable code

Full details: PRINCIPLES.md

📚 Documentation

Core Documentation

• Design Philosophy - Why this design (設計思想・意図)
• Commands Reference (English)
• Commands Reference (Japanese)
• Configuration Guide
• Japanese Configuration

Development Guides

• Principles Guide - Complete overview of all development principles
• Markdown Conventions - Markdown writing and reference rules

🤝 Contributing

Feel free to fork this repository and customize it for your needs. Pull requests for improvements are welcome!

📜 License

MIT License - Feel free to use and modify as needed.

👤 Author

thkt

---

This configuration enhances Claude AI's capabilities for systematic software development with a focus on quality, readability, and maintainability.