Git Prompts MCP Server

What is Git Prompts MCP Server?

Git Prompts MCP Server is a Model Context Protocol (MCP) server that offers several commands to generate prompts based on a Git repository's content. It integrates with MCP-compatible clients like the Zed Editor and Gemini CLI to provide Git-aware prompts and tools.

How to use Git Prompts MCP Server?

Prerequisites: Python 3.12+ and uv. For Zed Editor, add a configuration block to settings.json specifying the uvx command with arguments for repository path, excludes, and output format. For Gemini CLI, install the extension via gemini extension install https://github.com/ceshine/git-prompts-mcp-server.git --auto-update; commands and tools become available in Git repositories. The server relies on the current working directory being part of a Git repository. Environment variables GIT_REPOSITORY, GIT_EXCLUDES, and GIT_OUTPUT_FORMAT can be set for configuration.

Key features of Git Prompts MCP Server

• Generates Git-diff-based prompts for MCP-compatible clients.
• Provides commands like /git-diff, /generate-pr-desc, /git-cached-diff, /git-commit-messages, and /generate-commit-message.
• Offers equivalent MCP tools: git-diff, git-cached-diff, git-unstaged-diff, and git-commit-messages.
• Supports file exclusion patterns using pathlib.PurePath.
• Configurable output format: JSON (default) or text.
• Works as a Gemini CLI extension for easy installation and auto-updates.

Use cases of Git Prompts MCP Server

• Generate pull request descriptions from diff results between HEAD and an ancestor branch.
• Quickly populate staged or unstaged diff results for code review assistance.
• Get commit messages between two branches to summarize changes.
• Generate commit messages from staged changes, with potential issue highlights.

FAQ from Git Prompts MCP Server

What does Git Prompts MCP Server do compared to other Git MCP servers?

It focuses on generating prompts (commands and tools) based on Git diffs and commit history, drawing inspiration from the MarkItDown MCP server and the official Git MCP server.

What are the runtime requirements?

Python 3.12+ and uv are required. The server also needs a Git repository to operate.

Where does data live?

The server reads Git repository data from the local file system; no external data storage is used.

What transports/auth does the server use?

The server uses the Model Context Protocol (MCP) via stdio transport; no authentication is mentioned.

Are there any known limitations?

The extension relies on the current working directory being part of a Git repository; if not run from a Git project root, commands and tools may not function as expected.