github-mcp-tool
このMCPツールは、GitHubリポジトリ内でモデルのコンテキストを管理するための便利なツールです。モデルのバージョン、データセット、メトリクス、トレーニング設定などの重要な情報を文書化でき、リポジトリの作成やファイル操作も簡単に行えます。APIリクエストを通じて、ユーザー情報の取得やリポジトリの管理が可能です。
GitHubスター
1
ユーザー評価
未評価
お気に入り
0
閲覧数
33
フォーク
0
イシュー
1
README
🚀 MCP Tool for GitHub
Overview
The MCP (Model Context Protocol) Tool helps manage and track model context in GitHub repositories. It allows you to document important details like model versions, datasets, metrics, and training configurations.
📦 Overview
The MCP Tool for GitHub is designed to help manage:
- Repositories
- Files (create/update/delete)
- User information
- GitHub API requests
✨ Features
🧑💻 1. User Info
git_info(username: str)
- Fetch public GitHub user profile information.
📁 2. Repository Management
create_repo(repo_name: str, private: bool, description: str)
- Create a new repository (private/public).
delete_repo(username: str, repo_name: str)
- Delete a repository by username and repo name.
📄 3. File Operations in Repository
create_file(...)
- Add a new file to a repository.
- Content is automatically base64 encoded.
update_file(...)
- Update an existing file.
- Requires the file's SHA from GitHub.
delete_file(...)
- Delete a file.
- Requires SHA and commit message.
🔄 4. Request Utility
make_request(...)
- Handles all API interactions (
GET
,POST
,PUT
,DELETE
). - Authenticated via GitHub Token.
- Centralized method to perform various actions like:
- Create repositories
- Upload/update/delete files
- Fetch user/repo data
- Handles all API interactions (
🔐 5. Authentication & Setup
- Reads GitHub token securely from
.env
file. - Sets proper headers (
User-Agent
,Accept
,Authorization
). - Fully asynchronous requests using
httpx
.
🧰 Prerequisites
- ✅ Python 3.7+
- ✅ Git & uv installed
- ✅ GitHub Personal Access Token
📁 .env File Format
Create a .env
file in the root directory and add:
GITHUB_TOKEN=your_personal_access_token_here
🛠️ Setup & Usage
# Clone the Repo
git clone https://github.com/jalaj-pandey/github-mcp-tool.git
# Change Directory
cd github-mcp-tool
# Initialize virtual environment using uv
uv init
uv venv
.venv/Scripts/activate # Use `source ./venv/bin/activate` on Unix/Mac
# Add dependencies
uv add mcp[cli] httpx python-dotenv
# Start MCP dev server
mcp dev main.py