Llamit - AI-Powered Commit Messages

✨ Fully vibecoded - This project was entirely developed using AI assistance, showcasing the power of AI-driven development.

Generate semantic commit messages using Ollama or any OpenAI-compatible endpoint.

Run it locally with Ollama, or point it at the API gateway/model endpoint you already use.

Features

• 🚀 Generate commit messages instantly from staged changes
• 🔌 Provider-flexible - works with Ollama and OpenAI-compatible APIs
• 📝 Conventional Commits - follows standard commit message format
• ⚡ Fast - powered by a lightweight Go CLI
• 🎨 VS Code integration - seamless SCM toolbar button

Prerequisites

• Either Ollama running locally or an OpenAI-compatible API endpoint
• A compatible model name for the selected endpoint (default: qwen2.5-coder:7b)
• VS Code 1.85.0 or higher

Installation

Option 1: From VS Code Marketplace (Recommended)

1. Open VS Code
2. Go to Extensions view (Ctrl+Shift+X / Cmd+Shift+X)
3. Search for "Llamit"
4. Click Install

Option 2: Build from Source

# Clone the repository
git clone https://github.com/crstian19/llamit.git
cd llamit

# Build the Go CLI
cd go-cli
go build -o cli main.go

# Build the VS Code extension
cd ../vscode-extension
npm install
npm run compile

# Package the extension
npx vsce package
# Install the generated .vsix file in VS Code

Usage

1. Start your model endpoint

   • For Ollama:

   ollama serve
   

   • For OpenAI-compatible APIs: make sure you have the full chat/completions URL and, if required, an API key

2. Stage your changes: Use Git to stage the files you want to commit

   git add .
   

3. Generate commit message:

   • Click the ✨ Llamit button in the Source Control toolbar, or
   • Open Command Palette (Ctrl+Shift+P / Cmd+Shift+P)
   • Run: Llamit: Generate Commit Message

4. Review and commit: The generated message appears in the commit input box. Review it and commit!

Configuration

You can customize Llamit in VS Code settings:

{
  "llamit.apiType": "ollama",
  "llamit.apiUrl": "http://localhost:11434/api/generate",
  "llamit.model": "qwen2.5-coder:7b",
  "llamit.commitFormat": "conventional",
  "llamit.customFormat": ""
}

Settings

• llamit.apiType: Request/response compatibility mode (default: ollama)
  • Available values: ollama, openai-compatible
• llamit.apiUrl: Full API URL to call
  • Ollama example: http://localhost:11434/api/generate
  • OpenAI-compatible example: https://api.openai.com/v1/chat/completions
• llamit.apiKey: Optional API key for OpenAI-compatible endpoints
  • If empty, Llamit falls back to LLAMIT_API_KEY and then OPENAI_API_KEY
• llamit.ollamaUrl: Deprecated legacy setting kept for backward compatibility
• llamit.model: The model to use for generation (default: qwen2.5-coder:7b)
• llamit.commitFormat: The commit message format to use (default: conventional)
  • Available formats: conventional, angular, gitmoji, karma, semantic, google, custom
• llamit.customFormat: Custom format template (only used when commitFormat is set to custom)
• llamit.keepAlive: Ollama-only keep-alive control
• Advanced sampling settings are passed through fully for Ollama and partially for OpenAI-compatible APIs where the field exists (temperature, topP, numPredict, seed, stop)

Migrating from llamit.ollamaUrl

llamit.ollamaUrl is deprecated.

If you already have:

{
  "llamit.ollamaUrl": "http://localhost:11434/api/generate"
}

migrate to:

{
  "llamit.apiType": "ollama",
  "llamit.apiUrl": "http://localhost:11434/api/generate"
}

The extension includes a one-time migration prompt for users still relying on llamit.ollamaUrl.

Example: OpenAI-Compatible Endpoint

{
  "llamit.apiType": "openai-compatible",
  "llamit.apiUrl": "https://api.openai.com/v1/chat/completions",
  "llamit.model": "gpt-4o-mini",
  "llamit.apiKey": "",
  "llamit.commitFormat": "conventional"
}

Commit Message Formats

Llamit supports multiple commit message formats to match your team's conventions:

Conventional Commits (Default)

feat(auth): add user login functionality

Implements OAuth2 authentication flow

Angular

feat(core): implement user authentication

- Add login service
- Add auth guard
- Update routing

Closes [#123](https://github.com/crstian19/llamit/issues/123)

Gitmoji

✨ feat(api): add new endpoint for user profiles

Implements GET /api/users/:id endpoint

Karma

feat(ui): add dark mode toggle

Implements theme switching functionality

Semantic

feat: implement user authentication system

Complete OAuth2 integration with JWT tokens

Google

Add user authentication system

Implements a complete authentication flow using OAuth2 and JWT tokens.
Includes login, logout, and token refresh functionality.

Custom Format

Set llamit.commitFormat to custom and provide your own template in llamit.customFormat:

{
  "llamit.commitFormat": "custom",
  "llamit.customFormat": "Generate a simple commit message:\n<action>: <description>\n\nRules:\n1. Keep it under 50 characters\n2. Use imperative mood"
}

Recommended Models

Any Ollama model works, but these are optimized for code:

• qwen2.5-coder:7b - Great balance of quality and speed (default)
• qwen2.5-coder:14b - Better quality, slower
• codellama:13b - Good alternative
• deepseek-coder:6.7b - Fast and efficient

Architecture

Llamit consists of two components:

1. Go CLI (go-cli/)

A standalone command-line tool that:

• Reads git diffs from stdin
• Sends them to the configured LLM endpoint with a prompt template
• Returns a formatted commit message
• Implements retry logic with exponential backoff
• Handles errors gracefully

2. VS Code Extension (vscode-extension/)

A TypeScript extension that:

• Integrates with VS Code's Source Control view
• Executes git diff --cached to get staged changes
• Spawns the Go CLI as a subprocess
• Populates the commit message box with the result

Development

Running Tests

Go CLI:

cd go-cli
go test -v              # All tests
go test -v -short       # Unit tests only (skip integration)

VS Code Extension:

cd vscode-extension
npm run test:unit       # Fast unit tests
npm test               # Full integration tests

File Structure

llamit/
├── go-cli/              # Go CLI binary
│   ├── main.go          # Core logic
│   ├── main_test.go     # Comprehensive tests
│   └── go.mod           # Go module file
├── vscode-extension/    # VS Code extension
│   ├── src/
│   │   ├── extension.ts # Extension entry point
│   │   └── test/        # Unit and integration tests
│   ├── package.json     # Extension manifest
│   └── tsconfig.json    # TypeScript config
└── CLAUDE.md           # AI assistant documentation

Testing

Both components have comprehensive test coverage:

• Go CLI: 6 test cases covering success, errors, retries, and integration
• VS Code Extension: Unit tests + integration tests for all core functions

See CLAUDE.md for detailed testing information.

Telemetry

Llamit collects minimal anonymous telemetry for extension install and update events only.

• No code, diffs, commit messages, file paths, or repository names are collected
• No personal identifiers are collected
• Llamit respects VS Code's global telemetry settings

You can disable Llamit-specific telemetry with:

{
  "llamit.telemetry.enabled": false
}

See USAGE_DATA.md for the full policy.

Contributing

Contributions are welcome! This project was vibecoded, but that doesn't mean it can't be improved by humans too 😊

1. Fork the repository
2. Create a feature branch: git checkout -b feature/amazing-feature
3. Commit your changes: git commit -m 'feat: add amazing feature'
4. Push to the branch: git push origin feature/amazing-feature
5. Open a Pull Request

License

MIT License - see LICENSE file for details

Acknowledgments

• Built with Claude and other AI coding tools
• Powered by Ollama and OpenAI-compatible LLM backends
• Inspired by the need for better commit messages everywhere

Made with 🤖 and ✨ through vibecoding