What is Adk Github Mcp

The GitHub ADK Agent is a tool built with the Agent Development Kit (ADK) designed to interact with the GitHub remote MCP server. It provides users with a natural language interface to query information about repositories, users, and other GitHub data using advanced capabilities from Google’s Gemini model.

Use cases

The agent can be used for various purposes, such as discovering popular repositories based on specific topics, retrieving details about open issues in a repository, or acquiring recent commit information. Users can leverage these capabilities to gain insights and streamline their interactions with GitHub’s vast dataset.

How to use

To use the GitHub ADK Agent, clone the repository from GitHub, obtain necessary API keys from Google AI Studio and GitHub, and set them as environment variables. After the setup, run the agent’s development UI using the command ‘uv run adk web’, which starts a local web server for interacting with the agent via a web browser.

Key features

Key features of the GitHub ADK Agent include natural language processing capabilities powered by Google’s Gemini model, integration with the GitHub API, and support for various queries regarding repositories and users. This makes it a versatile tool for developers and researchers alike.

Where to use

The agent can be utilized in a variety of environments, including local development setups for individual developers, educational settings for teaching about GitHub API interactions, or any scenario where users need simplified access to GitHub’s extensive data through natural language queries.

GitHub ADK Agent

This repository contains a simple agent built with Agent Development Kit (ADK).
The agent is designed to interact with the new GitHub remote MCP server,
allowing you to ask questions and get information about GitHub repositories, users, and more.

Overview

The agent leverages the power of Google’s Gemini model to understand and respond to your queries.
It uses the MCPToolset from the Google ADK to connect to the GitHub remote MCP server,
providing a natural language interface to GitHub’s vast dataset.

Pre-requisites

Before you begin, ensure you have the following installed on your system:

• Python 3.10+
• uv: A fast Python package installer and resolver. You can install it by following the official instructions: https://github.com/astral-sh/uv

Setup

Follow these steps to set up the project locally.

1. Clone the Repository

First, clone the repository to your local machine and navigate into the project directory:

git clone https://github.com/jackwotherspoon/adk-github-mcp.git
cd adk-github-mcp

2. Get API Keys

This project requires two API keys to function:

Google AI Studio API Key

You’ll need a Google AI Studio API key to use the Gemini model.

1. Go to https://aistudio.google.com/apikey.
2. Click on “Create API key”.
3. Copy the generated API key.
4. Set it as an environment variable named GOOGLE_API_KEY:

export GOOGLE_API_KEY="YOUR_GOOGLE_API_KEY"

GitHub Personal Access Token (PAT)

To authenticate with the GitHub MCP server, you need a GitHub Personal Access Token.

1. Go to your GitHub Developer settings.
2. Click on “Personal access tokens” -> “Tokens (classic)”.
3. Click “Generate new token” -> “Generate new token (classic)”.
4. Give your token a descriptive name.
5. Set an expiration date for your token.
6. Important: For security, grant your token the most limited scopes necessary.
   For read-only access to repositories, the repo:status, public_repo, and
   read:user scopes are often sufficient. Avoid granting full repo or
   admin permissions unless absolutely necessary.
7. Click “Generate token”.
8. Copy the generated token.
9. Set it as an environment variable named GITHUB_PERSONAL_ACCESS_TOKEN:

export GITHUB_PERSONAL_ACCESS_TOKEN="YOUR_GITHUB_PERSONAL_ACCESS_TOKEN"

Running the Agent

Once you have completed the setup, you can run the agent’s development UI.

uv run adk web

This will start a local web server, and you can interact with the agent through your browser.

Example Queries

Here are some example queries you can try to test the agent’s capabilities:

• “Find the top 5 most starred repositories for the topic ‘python’.”
• “List the open issues for the ‘tensorflow/tensorflow’ repository.”
• “What are the latest commits in the ‘google/adk-python’ repository?”

These examples showcase how you can use natural language to query the GitHub MCP server through the agent.

Feel free to experiment with your own questions!