Claude Code Starter

A starter template for setting up Claude Code in any project.

This Repository

This is a template repo - clone it and run ./setup.sh to configure Claude Code for your project.

Code Style

When working in this repo:

• Write clear, readable code
• Use descriptive names
• Keep functions small and focused
• Follow existing patterns in the codebase

Security

• Never commit secrets or credentials
• Validate all external input
• Use environment variables for sensitive config
• .env and .env.* files are blocked from reading (see .claude/settings.json)
• See .claude/rules/security-model.md for full security documentation

Commit Review

Before committing changes, always:

1. Show the user what files are being committed
2. Briefly explain the key changes
3. Ask for confirmation before proceeding

This prevents "vibe coding" - blindly committing AI-generated changes without review.

Quick Commands

Simple workflow commands invoked with /<command> [arguments]:

• /onboard [area]: Quick project orientation and overview
• /pr-summary: Generate PR description from current changes
• /status: Check git state and recent activity

Available Skills

Skills are invoked automatically when your request matches their description:

• code-review: Review code changes for quality and security
• explain-code: Explain how code works
• generate-tests: Generate comprehensive tests
• refactor-code: Improve code without changing behavior
• review-mr: Review merge/pull requests
• install-precommit: Install the pre-commit review hook
• wiggum: Start autonomous implementation loop from spec/PRD (/wiggum)
• refresh-claude: Update CLAUDE.md with recent changes (/refresh-claude)
• changelog-writer: Maintain CHANGELOG.md with categorized entries (/changelog-writer)
• release-checklist: Final quality gate before shipping (/release-checklist)
• risk-register: Document risks for auth/data/migration changes (/risk-register)

Available Agents

Use these specialized agents for focused tasks:

• researcher: Explore and understand code (read-only)
• code-reviewer: Thorough code reviews
• code-simplifier: Simplify code for clarity and maintainability
• test-writer: Generate comprehensive tests
• documentation-writer: Generate and update documentation for code changes
• adr-writer: Create Architecture Decision Records for significant decisions

Wiggum Skill

The wiggum skill (inspired by the Ralph Wiggum technique) takes a specification or PRD and autonomously implements it to completion. It coordinates with specialized agents and enforces production-ready quality gates:

Workflow:

1. Plans first - Enters plan mode, identifies commands from CLAUDE.md, gets user approval
2. Implements iteratively in small chunks (blast-radius awareness)
3. Consults researcher when stuck
4. Documents decisions with adr-writer for significant choices
5. Validates dependencies - Checklist for any new packages (license, security, maintenance)
6. Uses test-writer for comprehensive test coverage
7. Gets code-reviewer feedback (includes security checklist)
8. Applies code-simplifier for clarity
9. Updates documentation with documentation-writer
10. Runs command gates per chunk (test, lint, typecheck)
11. Commits incrementally after each chunk passes all gates
12. Maintains changelog with changelog-writer
13. Final verification - All command gates, smoke test, production hygiene
14. Only finishes when ALL gates pass AND all agents approve

Invoke via: /wiggum "implement feature X per this spec..."

Project Structure

.claude/
├── settings.json              # Permissions, hooks
├── commands/                  # Quick workflow commands
│   ├── onboard.md
│   ├── pr-summary.md
│   └── status.md
├── skills/                    # Custom skills (YAML frontmatter + instructions)
│   ├── code-review/SKILL.md
│   ├── explain-code/SKILL.md
│   ├── generate-tests/SKILL.md
│   ├── refactor-code/SKILL.md
│   ├── review-mr/SKILL.md
│   ├── install-precommit/SKILL.md
│   ├── refresh-claude/SKILL.md
│   ├── wiggum/SKILL.md
│   ├── changelog-writer/SKILL.md
│   ├── release-checklist/SKILL.md
│   └── risk-register/SKILL.md
├── agents/                    # Specialized subagents
│   ├── researcher.md
│   ├── code-reviewer.md
│   ├── code-simplifier.md
│   ├── test-writer.md
│   ├── documentation-writer.md
│   └── adr-writer.md
├── hooks/                     # Automation scripts
│   ├── validate-bash.sh       # Pre-command validation
│   ├── auto-format.sh         # Post-edit formatting
│   └── pre-commit-review.sh   # Git pre-commit hook
└── rules/                     # Documentation (not auto-loaded)
    ├── code-style.md
    ├── git.md
    ├── quality-gates.md
    ├── security.md
    ├── security-model.md
    └── testing.md
stacks/                        # Stack-specific presets (starter repo only)
├── typescript/
├── python/
├── go/
├── rust/
├── ruby/
└── elixir/

Commands

./setup.sh    # Interactive setup for any project

For Users

After cloning, replace this CLAUDE.md with your project-specific version. See CLAUDE.template.md for a starting point.