This MCP server implements AI-assisted code quality analysis for iOS projects. When run without CLI arguments, it starts an MCP server that listens on localhost:3000 and responds to Model Context Protocol requests with structured, machine-readable analysis reports. The server integrates with your CI or AI assistants to provide actionable feedback about iOS test results, SwiftLint findings, and build diagnostics. If you prefer a direct CLI workflow, you can still invoke the bundled CLI via npx to run tests or lint locally, while the MCP server mode is designed for AI-assisted conversations and interactive debugging.

To use with an AI assistant, point the assistant at http://localhost:3000 and ensure it is configured to send MCP requests to that endpoint. The server returns contextual recommendations, file/line references, and potential next steps to improve code quality and test reliability. The CLI remains available for developers who want quick test runs, lint checks, or ad-hoc analysis without starting the MCP server.

Prerequisites:

• Node.js 18+ installed on your machine
• npm installed with Node.js
• Xcode and iOS tooling if you plan to run actual tests locally

Setup steps:

1. Clone the repository (or prepare your project as described in the MCP docs): git clone https://github.com/a-25/ios-mcp-code-quality-server.git cd ios-mcp-code-quality-server

2. Install dependencies: npm install

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

4. Start the MCP server (default MCP mode): npm start

   or explicitly as server mode

   node dist/index.js server

5. Verify the server is running: Open http://localhost:3000 in your browser or send a test MCP request from your AI assistant configuration.

Notes and tips:

• The MCP server runs by default on port 3000; you can override this with the PORT environment variable (e.g., PORT=3001).
• For debugging and verbose output, set LOG_LEVEL to debug.
• The server provides structured MCP responses suitable for AI assistants, including file locations and line numbers for issues found during analysis.
• The CLI (invoked via npx) supports commands like test and lint, enabling quick local workflows without starting the server.
• If you plan to integrate with an AI assistant, ensure your assistant sends MCP-compatible requests to the configured URL and handles the JSON responses accordingly.
• While SwiftLint is optional, having it installed improves code quality feedback when using the CLI lint command.