Optical Context MCP

Optical Context MCP is an MCP server built to help AI agents work efficiently with long, visually complex PDF documents. It reads local PDF files, uses Mistral OCR to extract both text and images, then recomposes that content into a smaller set of dense packed images which preserve original visual groupings. Agents (such as Claude or Windsurf) can then retrieve only the packed images they need for downstream processing. Ideal for operating manuals, product catalogs, or scanned handbooks, this MCP enables semantic, structured access to visually rich documents where plain OCR text isn’t enough.