The GitHub MCP server gives AI models structured access to repositories, pull requests, issues and code search via the Model Context Protocol. It is hosted by GitHub at api.githubcopilot.com/mcp/ and authenticates with a standard GitHub personal access token — no infrastructure to run.
https://api.githubcopilot.com/mcp/
Claude Sonnet 4.5
MCP Playground runs 30+ models on the same workflow: switch anytime, or use Compare mode to run several in parallel and balance quality vs. cost.
GitHub personal access token (fine-grained or classic) with the repo scopes you need. Generate one at github.com/settings/tokens.
How models use it and what it is built for.
GitHub publishes a hosted MCP server (api.githubcopilot.com/mcp/) that exposes the GitHub REST and GraphQL APIs as MCP tools. Models can list repos, read files at a ref, search code across an org, fetch and comment on pull requests, triage issues, inspect Actions workflows and more. You authenticate with a fine-grained or classic personal access token scoped to the repos you want the agent to touch. Because it is hosted, there is nothing to install — paste the URL and a token into any MCP client to start.
Typical tools an AI model can call. Exact names vary by version.
Copy any of these into MCP Agent Studio after connecting.
List open pull requests on this repo and flag any that look stale.
Summarise the last 5 merged PRs and what areas of the codebase they touched.
What files changed in PR #42 and are there any obvious risks?
Find issues labelled "bug" that have not been updated in 30 days.
This is not a single-model product: you get the same MCP connection with 30+ models (Claude, GPT, Gemini, DeepSeek, open-weight, and more), you can switch mid-conversation, and you can open Compare mode to run the same prompt against multiple models at once. The card above is a suggested starting point for this server — not the only choice.
Default pick for GitHub
Claude Sonnet 4.5
Claude Sonnet 4.5 leads on code-heavy reasoning and PR review. GPT-5.2 is a strong alternative for general repo Q&A; Haiku for cheap bulk triage.
Open MCP Agent Studio with the connection pre-filled. Add your token, pick any of 30+ models, and start chatting — no install required.
Try GitHub in Agent StudioCommon questions about connecting, scoping and using it safely.
It is an MCP-compatible server, hosted by GitHub at api.githubcopilot.com/mcp/, that exposes the GitHub API as tools an AI model can call. It lets models read repos, review pull requests, search code and manage issues without bespoke integrations.
The MCP server itself is free to use with any GitHub account. You authenticate with a personal access token. AI model usage is billed separately by your model provider (or by MCP Agent Studio if you use the in-browser credits).
Create a fine-grained personal access token with read access to the repos you want to query. Add write access only if you want the agent to create issues, comments or PRs. Avoid scoping more than you need.
Yes. Any MCP-compatible client works — Claude Desktop, Cursor, Cline, and browser tools like MCP Agent Studio. Paste the URL https://api.githubcopilot.com/mcp/ and your token into the client.
Copilot Chat is a closed product tied to GitHub-hosted models. The MCP server is an open protocol you can drive from any model — Claude, GPT, Gemini, DeepSeek — and it returns structured tool results, not natural-language answers.
Notion
Search, summarise and edit your Notion workspace with AI.
Vercel
List deployments, read logs, manage env keys and roll back from natural language.
Supabase
Manage Supabase projects, run SQL and inspect schemas with AI.
Cloudflare
Drive Workers, DNS, R2, D1 and the rest of the Cloudflare API with AI.