ko1ynnky/github-actions-mcp-server

GitHub Actions Model Context Protocol Server

Summary

GitHub Actions MCP Server is a TypeScript Model Context Protocol server that exposes the GitHub Actions API to AI coding assistants such as Claude Desktop, Codeium, and Windsurf. It provides nine tools for listing, inspecting, triggering, canceling, and rerunning workflows on a repository. The README carries an archive notice indicating the project will be retired as the official GitHub MCP server gains Actions support.

What it is useful for

This server is useful when an engineer wants a chat-based assistant to drive GitHub Actions without leaving the editor. Concretely, an assistant can call list_workflow_runs filtered by branch, actor, or status; fetch failed run details via get_workflow_run_jobs; re-run a build with rerun_workflow; or dispatch a deployment with trigger_workflow, passing workflowId, ref, and an inputs object. get_workflow_usage returns usage statistics including billable minutes, which is helpful for cost checks.

How engineers can use it

The README documents a local build. On Unix/Linux/macOS:

git clone https://github.com/ko1ynnky/github-actions-mcp-server.git
cd github-actions-mcp-server
npm install
npm run build

On Windows it specifies npm run build:win or the bundled run-server.bat. Listed prerequisites are Node.js v14 or higher and npm v6 or higher. The build produces dist/index.js, which the MCP client launches through node. For Claude Desktop, the README shows adding an mcpServers block to claude_desktop_config.json that runs node <path>/dist/index.js with GITHUB_PERSONAL_ACCESS_TOKEN in env. Codeium and Windsurf use the same shape, configured under ~/.codeium/windsurf/mcp_config.json on Unix or %USERPROFILE%\.codeium\windsurf\mcp_config.json on Windows. A documented usage example invokes triggerWorkflow({ owner, repo, workflowId: "ci.yml", ref: "main", inputs: { environment: "production" } }).

The README notes that authentication requires a GitHub token with sufficient scopes and that rate limiting, timeouts, and strict URL validation are enforced server-side. Given the archive notice and upcoming official Actions coverage, engineers should treat this as suitable for experimentation or local prototyping rather than long-term production adoption; full setup details and client-specific troubleshooting live in the project README.