Compresses OCR-heavy PDFs into dense packed images, enabling AI agents to efficiently process long visual documents.
Sponsored
Optical Context is an MCP server designed to streamline the processing of large, visually structured PDFs for AI agent workflows. It reads local PDFs, performs OCR using Mistral to extract page markdown and embedded images, then intelligently recomposes this content into dense PNGs. These packed images retain crucial visual grouping and structure, providing a more effective intermediate format for vision-capable agents compared to raw OCR text. The tool stores job artifacts and a manifest, allowing agents to selectively retrieve only the packed images they require, optimizing retrieval and processing for complex documents like operating manuals, scanned handbooks, and product catalogs.
Características Principales
01Packs extracted content into dense PNGs, preserving visual grouping.
02Extracts page markdown and embedded images using Mistral OCR.
030 GitHub stars
04Enables agents to fetch specific packed images via MCP.
05Stores job manifests and temporary artifacts for efficient follow-up retrieval.
06Provides MCP tools for PDF compression, manifest retrieval, and image fetching.
Casos de Uso
01Converting product catalogs and PDF slide decks into agent-friendly visual formats.
02Compressing visually structured, OCR-heavy documents for efficient AI processing.
03Processing operating manuals and scanned handbooks for AI agent analysis.
