GitHub radar
doc7: Any Document Into AI-Ready Markdown
Open-source cross-platform CLI that converts PDFs, Office files, scans, screenshots, charts, formulas, and diagrams into clean Markdown using any local or remote OpenAI-compatible vision model — no cloud OCR subscription or per-page fee required.
doc7 converts PDFs, Word and PowerPoint files, scans, screenshots, charts, diagrams, and formulas into Markdown through any OpenAI-compatible vision model, including local deployments via LM Studio or Ollama. Unlike classic OCR tools, it treats each page as an image and asks the model to recover the full semantic structure — tables, equations, figure captions, and cross-references — rather than just extracting character sequences. In a benchmark on raster-only pages from 'Attention Is All You Need', doc7 recovered 15 of 15 checked facts, while MarkItDown with its OCR plugin recovered 9 and Docling recovered 3. It is attracting attention because document conversion is a persistent bottleneck for anyone building RAG or agent pipelines around complex real-world files.
Why a vibe-coder should care
If your project works with contracts, research papers, invoices, or any document that contains charts or scanned images, doc7 turns those files into text your AI agent can actually read, quote, and reason over — running entirely on your own machine using a free local model, with no data sent anywhere.
▌ More finds