Skip to main content

An MCP server for gitingest

Project description

Gitingest MCP server

An MCP server for gitingest that provides access to Git repository analysis through the Model Context Protocol (MCP). This server leverages the gitingest library to analyze Git repositories and make their content available in a format optimized for LLMs.

Overview

This MCP server provides a single unified tool for accessing Git repository data. It automatically handles repository ingestion as needed, so users can immediately query repository content without an explicit ingestion step.

Tool: gitingest

The server provides a single tool called gitingest that can be used to analyze Git repositories. The tool accepts the following parameters:

  • repo_uri (required): URL or local path to the Git repository
  • resource_type: Type of data to retrieve (summary, tree, content, or all). Default is summary.
  • max_file_size: Maximum file size in bytes to include in the analysis. Default is 10MB.
  • include_patterns: Comma-separated patterns of files to include in the analysis.
  • exclude_patterns: Comma-separated patterns of files to exclude from the analysis.
  • branch: Specific branch to analyze.
  • output: File path to save the output to.

Resource Types and Large Repositories

For large repositories, it's recommended to first request only the summary (which is the default). After ingestion, you can access more detailed information through the resources:

  • Use the tree resource to explore the repository structure
  • Use the content resource to access the full content (if not too large)

If the repository is too large, consider using include_patterns and/or exclude_patterns to limit the scope of the ingestion.

Automatic Ingestion and Simple Caching

The server will automatically ingest repositories on demand. You don't need to call a separate ingestion function before querying.

Results are cached in memory during the server's runtime, so subsequent requests for the same repository will be faster. However, the cache is cleared when the server restarts.

Once a repository is ingested, you can access its data either by calling the gitingest tool again or by using the resources interface.

New: Context Truncation with max_tokens

The gitingest MCP tool now supports an optional max_tokens argument. If provided, the response will be truncated to at most max_tokens tokens (1 token = 4 characters). This helps LLMs manage context size more effectively.

Example usage:

{
  "repo_uri": "https://github.com/cyclotruc/gitingest",
  "resource_type": "content",
  "max_tokens": 2048
}

MCP Server Configuration

To use this MCP server, add the following to your MCP config:

"mcpServers": {
  "trelis-gitingest-mcp": {
    "command": "uvx",
    "args": [
      "git+https://github.com/TrelisResearch/gitingest-mcp"
    ]
  }
}
  • The short config above works as long as the entry point and server name are both trelis-gitingest-mcp (see pyproject.toml and src/gitingest_mcp/server.py).
  • If you rename the repo or entry point, you may need to use the more verbose config with --from and --stdio.

Dependency Update

This MCP server now uses the official cyclotruc/gitingest repository for ingestion, ensuring you get the latest features and fixes.

Quickstart

[!TIP] The PyPI package trelis-gitingest-mcp does not yet support nested directories because that is supported by a branch that has not yet been merged in the gitingest library (and PyPI won't allow me to upload a package with a dependency on a branch). For this reason, using the github link to TrelisResearch/gitingest-mcp is recommended for now.

Install

Claude Desktop

On MacOS: ~/Library/Application\ Support/Claude/claude_desktop_config.json On Windows: %APPDATA%/Claude/claude_desktop_config.json

Published Servers Configuration

WARNING: Windsurf does not support passing in a github url. Use trelis-gitingest-mcp instead.

"mcpServers": {
  "trelis-gitingest-mcp": {
    "command": "uvx",
    "args": [
      "git+https://github.com/TrelisResearch/gitingest-mcp"
    ]
  }
}

Development/Unpublished Servers Configuration

"mcpServers": {
  "trelis-gitingest-mcp": {
    "command": "uv",
    "args": [
      "--directory",
      "/Users/your-username/trelis-gitingest-mcp",
      "run",
      "trelis-gitingest-mcp"
    ]
  }
}

Development

Building and Publishing

To prepare the package for distribution:

  1. Sync dependencies and update lockfile:
uv sync
  1. Build package distributions:
uv build

This will create source and wheel distributions in the dist/ directory.

  1. Publish to PyPI:
uv publish

Note: You'll need to set PyPI credentials via environment variables or command flags:

  • Token: --token or UV_PUBLISH_TOKEN
  • Or username/password: --username/UV_PUBLISH_USERNAME and --password/UV_PUBLISH_PASSWORD

Debugging

Since MCP servers run over stdio, debugging can be challenging. For the best debugging experience, we strongly recommend using the MCP Inspector.

Examples & MCP Inspector

Get everything for a repository:

{
  "repo_uri": "https://github.com/user/repo"
}

Get just the summary:

{
  "repo_uri": "https://github.com/user/repo",
  "resource_type": "summary"
}

Get the file tree:

{
  "repo_uri": "https://github.com/user/repo",
  "resource_type": "tree"
}

Get the full content:

{
  "repo_uri": "https://github.com/user/repo",
  "resource_type": "content"
}

Custom ingestion parameters:

{
  "repo_uri": "https://github.com/user/repo",
  "branch": "dev",
  "exclude_patterns": "*.md,tests/*",
  "max_file_size": 5242880
}

Using include patterns to focus on specific files:

{
  "repo_uri": "https://github.com/user/repo",
  "include_patterns": "src/*.py,bin/*"
}

Note: The patterns src/*.py and bin/* will only match Python files directly in the src directory and any files directly in the bin directory, not in subdirectories.

Save output to a file:

{
  "repo_uri": "https://github.com/user/repo",
  "resource_type": "all",
  "output": "/path/to/output/file.txt"
}

You can launch the MCP Inspector via npm with this command:

npx @modelcontextprotocol/inspector uv --directory /Users/your-username/trelis-gitingest-mcp run trelis-gitingest-mcp

or using uvx for the mcp server:

npx @modelcontextprotocol/inspector uvx https://github.com/TrelisResearch/gitingest-mcp.git

Upon launching, the Inspector will display a URL that you can access in your browser to begin debugging.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

trelis_gitingest_mcp-1.0.0.tar.gz (29.7 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

trelis_gitingest_mcp-1.0.0-py3-none-any.whl (11.7 kB view details)

Uploaded Python 3

File details

Details for the file trelis_gitingest_mcp-1.0.0.tar.gz.

File metadata

File hashes

Hashes for trelis_gitingest_mcp-1.0.0.tar.gz
Algorithm Hash digest
SHA256 5d4aeb3120e5eb0ab82c5ed18ec6ab9c170b9a519ab07f16eca7a9519d94f064
MD5 ed2114ee0b94cf964be7f45844c50302
BLAKE2b-256 ff12e399661713c9fc21503b44f9c9b230e796681f0fe2806ab7ceba3702767e

See more details on using hashes here.

File details

Details for the file trelis_gitingest_mcp-1.0.0-py3-none-any.whl.

File metadata

File hashes

Hashes for trelis_gitingest_mcp-1.0.0-py3-none-any.whl
Algorithm Hash digest
SHA256 df292070341d60ab4930271e13e3d8804b85da9dfe79197943f5def9b61c5066
MD5 90678feb009487ba9029330230abca82
BLAKE2b-256 8ffe90db8595eedf647c3b22bdc33d44ee24cfe1b54284460a8ed073e31aeca2

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page