Claude Code Configuration

A centralized repository for managing configuration files, prompts, and settings for Claude Code.

Purpose

This repository serves as a single source of truth for Claude Code configurations, enabling:

• Consistency: Maintain the same AI behavior and settings across all your projects
• Version Control: Track changes to your AI configurations over time
• Sharing: Easily share proven configurations with team members
• Portability: Quickly set up AI development environments on new machines
• Experimentation: Test different configurations without losing working setups

What's Included

Pre-configured Agents

• git-commit-manager: Analyzes code changes and generates semantic commit messages following conventional commit standards. Handles staging, committing, and ensures atomic commits with meaningful descriptions.

• project-documentation-keeper: Maintains your CLAUDE.md file by documenting architectural changes, new features, environment variables, and significant project developments. Proactively keeps documentation in sync with code.

• test-runner-fixer: Runs your test suite periodically and intelligently fixes failures while preserving test intent. Distinguishes between legitimate code changes and actual bugs, maintaining test quality and coverage.

• web-research-specialist: Researches current documentation, libraries, and best practices from the web. Provides up-to-date technical information with proper sources and implementation examples.

Other Features

• Custom Statusline: Enhanced statusline display showing current model and directory
• Project Instructions: CLAUDE.md configurations for context and behavior
• Custom prompts and templates: (Coming soon)
• Hooks: (Coming soon)

Statusline Display

This configuration includes a custom statusline that displays the current Claude model and working directory:

Prerequisites

Before using this configuration, ensure you have the following tools installed:

• jq: JSON processor required for statusline parsing

  # macOS with Homebrew
  brew install jq
  
  # Ubuntu/Debian
  sudo apt-get install jq
  
  # Other systems: https://github.com/jqlang/jq/releases

Installation

Quick Install (Recommended)

1. Clone this repository:

git clone <your-repo-url> ai-dev-tools-config
cd ai-dev-tools-config

2. Run the install script:

./install.sh

This will symlink the .claude directory to ~/.claude, making it your global Claude Code configuration.

Manual Installation

If you prefer manual setup:

# Clone the repository
git clone <your-repo-url> ~/.ai-dev-tools-config

# Create symlink to make it your global config
ln -sf ~/.ai-dev-tools-config/.claude ~/.claude

Usage

Customization for specific projects

• Override specific settings by creating local configuration files
• Use project-specific branches for different configuration sets
• Maintain a base configuration that works across most projects

Using Agents

Claude Code agents are specialized assistants that can be invoked using the Task tool. You can launch multiple agents concurrently to handle different aspects of your development workflow.

Example: Concurrent Agent Usage

When you've made significant code changes and want to ensure everything is properly documented, tested, and committed, you can launch multiple agents simultaneously:

# In Claude Code, you might say:
"I've just finished implementing the new payment processing feature.
Can you run the tests, update the documentation, and prepare a commit?"

# Claude Code will then:
# 1. Launch test-runner-fixer to run and fix any test failures
# 2. Launch project-documentation-keeper to update CLAUDE.md with the new feature
# 3. Launch git-commit-manager to create a semantic commit message
# All three agents work concurrently, saving time and ensuring consistency

Individual Agent Examples

# Research a new technology
"How do I implement WebSockets in Rails 8?"
# → Launches web-research-specialist

# Document a new feature
"I just added Redis caching with REDIS_URL environment variable"
# → Launches project-documentation-keeper

# Commit your changes
"Please commit these authentication changes"
# → Launches git-commit-manager

# Fix failing tests
"Can you run the test suite and fix any issues?"
# → Launches test-runner-fixer

Agents can also work together. For example, the test-runner-fixer might communicate with the project-documentation-keeper to understand feature requirements when fixing complex test failures.

Structure

.
├── .claude/              # Claude Code configurations
│   ├── agents/          # Pre-configured agents
│   │   ├── git-commit-manager.md
│   │   ├── project-documentation-keeper.md
│   │   ├── test-runner-fixer.md
│   │   └── web-research-specialist.md
│   └── CLAUDE.md        # Project context and instructions
└── README.md            # This file

Best Practices

• Keep sensitive information out of shared configurations
• Use descriptive commit messages when updating configurations
• Test configurations in isolated environments before applying broadly
• Document any project-specific customizations needed