GitHub Repository Intelligence

What GitHub Repository Intelligence MCP server does

GitHub Repository Intelligence MCP server gives MCP-compatible coding agents a read-only way to inspect GitHub repositories without cloning them locally. It returns repository metadata such as stars, forks, primary language, license, topics, and last push time. Agents can also inspect recent pull requests, contributor activity, file-change concentration, check results, and dependencies.

The service is intended for questions that require repository context rather than file editing. Examples include determining whether a project is still maintained, identifying its main maintainers, finding areas with recent churn, checking whether a branch is green, and reviewing declared dependencies.

How it works

The server runs on Apify and exposes a Streamable HTTP MCP endpoint. An MCP client sends tool calls to the endpoint with an Apify bearer token in the request headers. Each call identifies a repository by its owner and name; the tools then obtain current information from the GitHub REST API.

Six tools are available:

• get_repo_overview returns repository metadata and the last push date.
• list_recent_prs returns recent pull requests, including authors, states, and merge dates.
• get_contributors ranks contributors by commit count.
• get_hot_files identifies the most-modified files across the last requested number of commits.
• get_ci_status reports check runs and conclusions for a reference.
• get_dependencies parses package.json, requirements.txt, go.mod, Cargo.toml, and pyproject.toml.

The GitHub Repository Intelligence MCP server does not cache responses, so results reflect the data available from GitHub when each request runs. Every tool accepts an optional github_token argument. Without it, public-repository requests use GitHub’s unauthenticated limit of 60 requests per hour. A fine-grained personal access token provides a 5,000-request hourly limit and access to private repositories.

Setup and configuration

Configure an MCP client with the endpoint https://cg-nguyen--github-repo-intel-mcp.apify.actor/mcp and send the Apify token as an Authorization: Bearer <APIFY_TOKEN> header.

No local runtime, repository clone, Node or Python installation, or local server deployment is required. The hosted endpoint is shared across users, while authentication is handled through the Apify token. GitHub authentication can be supplied per tool call through github_token when higher limits or private repositories are needed.

Usage is priced at $0.005 per executed tool call, with the first 100 calls free per user.

Limitations and notes

The GitHub Repository Intelligence MCP server is read-only. It cannot create pull requests, comment on issues, or change repository state.

get_hot_files can inspect up to 200 commits and makes one GitHub API request per commit, making it the most rate-limit-intensive tool. Supply a GitHub token if using it frequently.

get_ci_status reads GitHub’s Checks API. It can report GitHub Actions, CodeQL, and third-party checks that publish through that API, but external CI systems without GitHub integration may produce zero check runs even when their builds pass. Responses include a note for this situation.