Pharaoh

This hooks your AI coding agent into a knowledge graph of your entire codebase, so it can query architectural context before writing or changing code. Instead of reading files blindly, Claude or Cursor can ask what depends on a function, what breaks if you refactor it, whether similar code already exists, or which functions are dead. Pharaoh maps your repo on every push via GitHub webhook, extracting function signatures, imports, and call graphs without storing source. It exposes operations for blast radius analysis, dependency tracing, and reachability checks from production entry points. Supports TypeScript and Python today. Useful when your agent keeps duplicating code, breaking shared modules, or proposing changes without understanding what they'll affect downstream.