GitHub MCP Agent MCP server
Commit Oct 24, 2026Allows an AI agent to explore GitHub repositories, inspect issues, pull requests, and repository activity through conversational queries.
Also known as abduIwahid/github-mcp-agent
- Stars
- 0
- Tools
- —
- Downloads, 30 days
- —
- Latest version
- —
- Last commit
- Oct 24, 2026
- License
- Apache-2.0
What is GitHub MCP Agent MCP?
GitHub MCP Agent allows an AI agent to explore GitHub repositories, inspect issues, pull requests, and repository activity through conversational queries.
Summary generated from the project's README
About GitHub MCP Agent MCP server
Exploring GitHub repositories through natural language conversation is the primary function of this tool. It allows users to inspect issues, pull requests, and general repository activity by asking questions in plain English. The system uses the Agno framework and Google Gemini 2.5 Flash to reason through repository data provided via the Model Context Protocol.
Developers run this application locally using Streamlit and Docker. To set it up, you must have Docker running, a GitHub personal access token with repo scope, and a Google Gemini API key. The official github-mcp-server runs inside a Docker container, ensuring your GitHub token stays on your machine and only communicates with GitHub. This setup provides a web interface where you can name a specific repository and begin querying its contents immediately.
What you can do with it
- 1Monitor repository discussionsAsk the agent to identify which issues are currently seeing active engagement from the community.
- 2Track pull request statusIdentify which pull requests are currently pending a review from maintainers.
- 3Summarize recent repository changesRequest a list of pull requests that have been merged into the codebase recently.
- 4Assess project healthGet a high-level overview of how active a specific repository has been lately.
Written from the project's README. Every use case is checked against a quote from it.
README
View on GitHubGitHub MCP Agent
Explore any GitHub repository in plain English. This agent connects to GitHub through the Model Context Protocol (MCP) — the same standard interface used by modern AI tooling — so it can inspect issues, pull requests and repository activity and answer questions about them conversationally.
What it does
- Ask about open issues, active discussions, and PRs that need review
- Get repository health and activity summaries without clicking through GitHub
- Runs the official GitHub MCP server locally in Docker — your token never leaves your machine except to GitHub
Tech stack
- Streamlit — web UI
- Agno + MCP tools — agent connected to the GitHub MCP server over stdio
- Google Gemini 2.5 Flash — reasoning model
- Docker — runs the official
ghcr.io/github/github-mcp-server
Run it locally
Prerequisites: Docker installed and running, a GitHub personal access token (repo scope), and a Google Gemini API key.
pip install -r requirements.txt
streamlit run github_agent.py
Paste both keys in the sidebar, name a repository in your question, and ask away.
Example questions
- "What issues are being actively discussed in facebook/react?"
- "Show me recent merged PRs in this repository."
- "Which PRs are waiting for review?"
Author
Abdul Wahid — AI/ML Developer Portfolio · GitHub · LinkedIn
Credits
Adapted from a template in the awesome-llm-apps collection by Shubham Saboo, and reworked to run on Gemini instead of OpenAI. Licensed under Apache-2.0 — see LICENSE.
GitHub MCP Agent MCP questions
How does the agent handle security and my GitHub token?
The agent runs the official GitHub MCP server locally in Docker, ensuring your token never leaves your machine except to go to GitHub.
What information can I ask the agent about a repository?
You can ask about open issues, active discussions, pull requests that need review, and general repository health or activity summaries.
What are the prerequisites to run this locally?
You need Docker installed and running, a GitHub personal access token with repo scope, and a Google Gemini API key.