Extract data from documents with an AI agent: text, tables, images and barcodes, run locally in Claude DeskTop, Cursor or VS Code through a Docker MCP server....pdf say, and what is in its table...its context window. Paste a PDF into a chat and the model works...