github-actions

The GitHub Actions MCP Server provides a programmable interface to manage and analyze GitHub Actions workflows. It exposes tools to list workflows, fetch workflow details, retrieve usage statistics, list and inspect workflow runs and their jobs, and trigger, cancel, or rerun workflow runs. Designed to work with AI coding assistants like Claude Desktop, Codeium, and Windsurf, you can plug this server into your preferred assistant using a standard MCP config. To use it, start the server in your environment with the required GitHub Personal Access Token and point your assistant’s MCP config to the server’s node entrypoint. The tools you’ll access will allow you to programmatically discover workflows in a repository, inspect run results and job details, and control workflow executions from your AI assistant or automation scripts.

Prerequisites:

• Node.js v14 or higher
• npm (comes with Node.js)

Installation steps:

1. Clone the repository: git clone https://github.com/ko1ynnky/github-actions-mcp-server.git
2. Install dependencies: cd github-actions-mcp-server npm install
3. Build the server (produces dist/index.js): npm run build
4. Run the server (example): node dist/index.js

Windows users can follow the dedicated build/run steps in the README, including npm run build:win or using the provided run-server batch file. Ensure your environment has a GitHub Personal Access Token with appropriate scopes and set it as GITHUB_PERSONAL_ACCESS_TOKEN when starting or configuring the MCP server.

Tips and considerations:

• You must provide a GitHub Personal Access Token (PAT) with sufficient scopes to access workflow data. Set it via the GITHUB_PERSONAL_ACCESS_TOKEN environment variable in your MCP config.
• The server is designed for compatibility with multiple AI assistants; use the example MCP config snippets for Claude Desktop, Codeium, or Windsurf as a guide.
• If you encounter rate limit issues, ensure your PAT has the required scopes and consider implementing proper PAT rotation and retry strategies in your integration.
• When building on Windows, use npm run build:win and the run-server.bat helper for a smoother setup.
• The MCP server exposes a range of endpoints (list_workflows, get_workflow, get_workflow_usage, list_workflow_runs, get_workflow_run, get_workflow_run_jobs, trigger_workflow, cancel_workflow_run, rerun_workflow); refer to the README for exact input parameters and return structures.
• For production deployments, consider containerizing the server and configuring environment variables securely.