MCP server that enables cloud models (like Gemini/Claude) to delegate coding tasks to a local llama.cpp server, preserving cloud usage limits through an AI-powered code review loop.
