MCP Git Commit Generator

MCP Git Commit Generator

Generate conventional commit messages from your staged git changes using Model Context Protocol (MCP).

✨ Features

• Automatic commit message generation based on staged git diffs.
• Supports Conventional Commits.
• MCP server with both stdio (default) and SSE transport options.
• Inspector UI for interactive inspection (via MCP Inspector).

📦 Requirements

• For Docker usage: Docker (for running the server in a container)
• For PyPI/uvx usage: Python >= 3.13.5 and uv (recommended) or pip
• Git (for version control)
• An MCP-compatible client (VS Code with MCP extension, Claude Desktop, Cursor, Windsurf, etc.)

🚀 Installation

You can install and use the MCP Git Commit Generator in multiple ways:

Option 1: Using uvx (Recommended)

The easiest way to use the package is with uvx, which automatically manages the virtual environment:

Option 2: Install from PyPI

Or with uv:

Option 3: Using Docker

Use the pre-built Docker image from GitHub Container Registry (no installation required):

🛠️ Available Tools

This MCP server provides the following tools to help you generate conventional commit messages:

generate_commit_message

Generates a conventional commit message based on your staged git changes.

Parameters:

• repo_path (string, optional): Path to the git repository. If omitted, uses the current directory.
• commit_type (string, optional): Conventional commit type (e.g., feat, fix, docs, style, refactor, perf, build, ci, test, chore, revert). If omitted, the type will be auto-detected.
• scope (string, optional): Scope of the change (e.g., file or module name). If omitted, the scope will be auto-detected based on changed files.

Usage:

1. Stage your changes: git add <files>
2. Use the tool through your MCP client to generate a commit message
3. The tool will analyze your staged changes and generate an appropriate conventional commit message

check_git_status

Checks the current git repository status, including staged, unstaged, and untracked files.

Parameters:

• repo_path (string, optional): Path to the git repository. If omitted, uses the current directory.

Usage:

Use this tool to get an overview of your current git repository state before generating commit messages.

🧩 MCP Client Configuration

Configure the MCP Git Commit Generator in your favorite MCP client. You have multiple options:

1. Using uvx (recommended - automatically manages dependencies)
2. Using Docker (no local Python installation required)
3. Using local Python installation (for development)

VS Code

Add one of the following configurations to your VS Code mcp.json file (usually located at .vscode/mcp.json in your workspace):

Using uvx (Recommended)

Using Docker

If you want to put the configuration in your user settings.json file, you can do so by adding:

Cursor

Add one of the following to your Cursor MCP configuration file (usually located at ~/.cursor/mcp.json):

Cursor with uvx (Recommended)

Cursor with Docker

Windsurf

Configure Windsurf with one of the following MCP server settings (usually located at ~/.codeium/windsurf/mcp_config.json):

Windsurf with uvx (Recommended)

Windsurf with Docker

Claude Desktop

Add one of the following to your Claude Desktop configuration file (usually located at ~/Library/Application Support/Claude/claude_desktop_config.json on macOS):

Claude Desktop with uvx (Recommended)

Claude Desktop with Docker

Note: The --mount option in Docker configurations allows the Docker container to access your home directory, enabling it to work with git repositories located anywhere in your file system. When using uvx or pip installations, this mounting is not needed as the tool runs directly on your system. Adjust the mount path if your repositories are located elsewhere when using Docker.

🚀 Quick Start Guide

1. Install the package using one of the methods above:
   • Recommended: uvx mcp-git-commit-generator (or configure in your MCP client)
   • Alternative: pip install mcp-git-commit-generator
   • Docker: Use the configurations above with Docker
2. Configure your MCP client using one of the configurations above
3. Stage some changes in a git repository:
   git add <files>
4. Use the tools through your MCP client:
   • Use check_git_status to see your current repository state
   • Use generate_commit_message to create a conventional commit message
5. Commit your changes with the generated message

👨‍💻 Developer Guidelines

The following sections are intended for developers who want to contribute to or modify the MCP Git Commit Generator.

Local Development Setup 🛠️

If you prefer not to use Docker for development, you can run the server locally:

Requirements:

• Python >= 3.13.5
• MCP CLI >= 1.10.1
• uv (for dependency management, optional but recommended)
• Node.js (for Inspector UI, optional)
• Python Debugger Extension (for debugging, optional)

Installation:

1. Clone the repository:
   git clone https://github.com/theoklitosBam7/mcp-git-commit-generator.git cd mcp-git-commit-generator
2. Prepare environment: There are two approaches to set up the environment for this project. You can choose either one based on your preference.

   Note: Reload VSCode or terminal to ensure the virtual environment python is used after creating the virtual environment.

   Approach	Steps
   Using uv	1. Create virtual environment: uv venv 2. Run VSCode Command "Python: Select Interpreter" and select the python from created virtual environment 3. Install dependencies (include dev dependencies): uv pip install -r pyproject.toml --group dev 4. Install mcp-git-commit-generator using the command: uv pip install -e ..
   Using pip	1. Create virtual environment: python -m venv .venv 2. Run VSCode Command "Python: Select Interpreter" and select the python from created virtual environment 3. Install dependencies: pip install -e .. 4. Install pip dev dependencies: pip install -r requirements-dev.txt.

3. (Optional) Install Inspector dependencies:
   cd inspector npm install

📦 Publishing to PyPI

The project includes an automated PyPI publishing workflow (.github/workflows/pypi-publish.yml) that:

• Triggers on: Tag pushes matching v*.*.* pattern, manual workflow dispatch, or pull requests to main
• Builds: Python package distributions using the build package
• Publishes: Automatically publishes to PyPI using trusted publishing (OIDC) when tags are pushed

To publish a new version:

1. Update the version in pyproject.toml
2. Create and push a git tag: git tag vX.Y.Z && git push origin vX.Y.Z
3. The workflow will automatically build and publish to PyPI

🐳 Building and Running with Docker

You can build and run the MCP Git Commit Generator using Docker. The provided Dockerfile uses a multi-stage build with uv for dependency management and runs the server as a non-root user for security.

Build the Docker image

Run the server in a container (default: stdio transport)

You can run the published image directly from GitHub Container Registry.

By default, the container runs:

If you want to use SSE transport (for Inspector UI or remote access), override the entrypoint or run manually:

The server will be available at http://localhost:3001 when using SSE.

🖥️ Running the Server Locally

To run locally (without Docker):

1. Set up your uv or Python environment as described in the Local Development Setup section.
2. From the project root, run:

You can specify other options, for example:

The server listens on 0.0.0.0:3001 by default when using SSE, or as specified by the options above.

Note:

• If you want to use the CLI entrypoint, ensure the package is installed and your environment is activated.
• Do not use positional arguments (e.g., python -m mcp_git_commit_generator sse); always use options like --transport sse.
• Available arguments with their values are:
  • --transport: Transport type (e.g., stdio (default), sse).
  • --host: Host to bind the server (default: 0.0.0.0).
  • --port: Port to bind the server (default: 3001).
  • -v, --verbose: Verbosity level (e.g., -v, -vv).

🔎 Start the Inspector UI

From the inspector directory:

The Inspector UI will be available at http://localhost:5173.

🗂️ Project Structure

⚙️ Advanced MCP Server Configuration for Development

The .vscode/mcp.json file configures how VS Code and related tools connect to your MCP Git Commit Generator server. This file defines available server transports and their connection details, making it easy to switch between different modes (stdio is default, SSE is optional) for development and debugging.

Example Development mcp.json

• mcp-git-commit-generator: Runs the server in a Docker container (default: stdio transport), using the published image.
• sse-mcp-git-commit-generator: Connects to the MCP server using Server-Sent Events (SSE) at http://localhost:3001/sse. Only useful if you run the server with --transport sse.
• stdio-mcp-git-commit-generator: Connects using standard input/output (stdio), running the server as a subprocess. This is the default and recommended for local development and debugging.
• uvx-mcp-git-commit-generator: Uses uvx to automatically install and run the package from PyPI.

🐞 Debugging the MCP Server

Notes:

• MCP Inspector is a visual developer tool for testing and debugging MCP servers.

  • All debugging modes support breakpoints, so you can add breakpoints to the tool implementation code.
  • You can test tool arguments directly in the Inspector UI: When using the Inspector, select a tool and provide arguments in the input fields to simulate real usage and debug argument handling.

⚙️ Default Ports and Customizations

💬 Feedback

If you have any feedback or suggestions, please open an issue on the MCP Git Commit Generator GitHub repository

📄 License

MIT License © 2025 Theoklitos Bampouris