GitHub radar
doc7: Convert any document to AI-friendly markdown
An open source CLI tool for converting PDFs, Office documents, scans, screenshots, graphs, equations, and diagrams to Markdown with any OpenAI compatible vision model (local or remote) without paying for a cloud OCR service or per-document fees.
doc7 leverages any OpenAI compatible vision model (e.g., LM Studio or Ollama running locally/remotely) to transform PDFs, MS Word/Powerpoint files, scanned documents, screenshots, charts, graphs, and equations into markdown. Rather than relying on conventional OCR techniques that try to read the text off of an image, doc7 takes an image-to-text approach by feeding the whole page to the model and asking it to reproduce all of the semantic content including tables, equations, figures, and citations. For example, on the Attention Is All You Need paper, doc7 was able to accurately capture all 15 pieces of information while MarkItDown+OCR only captured 9 and Docling captured 3. doc7 has been growing in popularity because many times we need to translate documents when building RAG applications or agents that interact with complex documents.
Why a vibe-coder should care
Using doc7, you can transform any type of document relevant to your project (e.g., contracts, research papers, invoices, etc.) into human-readable text so that your AI agent can access and reason over them locally instead of having to send this information to a server via a free local model.
▌ More finds