Built this to feed documents into LLM pipelines cleanly. Supports 15+ formats via a unified CLI, MCP, and Python API.