A self-hosted codebase indexing tool that runs a local LLM alongside your project to produce dense, AI-optimized reference documents. Scans every file, summarizes code using keyword-rich briefings, reconstructs database schemas from migrations, extracts WebSocket protocol surfaces, tracks flags and TODOs, detects sync drift between files, and generates a full project tree with dependency maps. Designed to fit an entire codebase into an AI assistant's context window without burning tokens on implementation details. Runs on Ollama. Works with Python, GDScript, JavaScript, and TypeScript.