mcp-stockfish is a Model Context Protocol server that bridges AI systems to the Stockfish chess engine. It exposes Stockfish to LLMs and other AI agents through MCP, enabling multiple concurrent sessions so several positions can be analyzed in parallel. Built on mark3labs/mcp-go, it leverages the MCP protocol to manage sessions, UCI state, and command execution, while providing robust error handling and a JSON-centric interface. The server supports the full UCI command set (uci, isready, position, go, stop, quit, etc.), and can run in stdio mode or HTTP mode for web-scale deployments. It also ships with Docker readiness for containerized deployment and is designed to prevent resource exhaustion with session timeouts and concurrency controls. The Quick Start demonstrates installation, usage, and a sample Claude Desktop integration. It emphasizes reliable operation, JSON-formatted responses, and straightforward integration with existing AI workflows.