This MCP server automates the creation and review of GitHub pull requests using AI-assisted analysis. It analyzes the current repository context, generates a conventional PR title, crafts a detailed or template-based description, and can automatically suggest and assign reviewers based on historical contributions. The unified workflow can be invoked via the create_pr_complete command, which orchestrates an AI-powered code review, PR description generation, GIF selection, and GitHub PR creation. You can also use the separate tools (review, generate_pr_title, generate_pr_description, suggest_reviewers, etc.) to compose PRs step-by-step or to inspect project context before creating a PR. The server is designed to work across languages and tech stacks, detecting TypeScript, i18n use, state management, styling, and testing frameworks to tailor recommendations and reviews to your project.

Prerequisites:

• Node.js v18+ installed on your machine
• A Git repository to operate within

Installation steps:

1. Clone the MCP server repository: git clone https://github.com/valentin-harrang/pr-mcp-server.git cd pr-mcp-server

2. Install dependencies: npm install

3. Build the project (if applicable): npm run build

4. Link the server globally for easy access (recommended): npm link

   This makes the pr-mcp-server command available globally

5. (Optional) Start the server locally to test: npm start

6. Verify integration with MCP Inspector (optional): npx @modelcontextprotocol/inspector pr-mcp-server

Tips and common considerations:

• Environment: Set GITHUB_TOKEN with appropriate scopes to enable repository access and PR creation.
• Run inside a Git repository to ensure branch/context detection works correctly.
• The server detects project context (language, frameworks, i18n, testing) to tailor AI reviews and PR content.
• If you encounter permission or token issues, double-check access rights and that the token is exported in the environment where the server runs.
• For best results, use the unified workflow create_pr_complete to leverage AI review, descriptive PR content, conventional titles, and automatic reviewer assignment.
• In Cursor or Claude Desktop workflows, reference the pr-mcp-server by its configured name (pr-mcp-server) in your mcp.json or integration settings.