anydoc

Use it with caution: anydoc has a clear role as a local document-to-Markdown preprocessing component, but the available evidence is dominated by launch posts and community amplification on X. It is not enough to establish broad independent testing or production readiness for critical workflows. The more accurate analogy is a parsing layer for document ingestion, not an AI agent, vector database, knowledge base, or document-management suite. Teams needing model reasoning, retrieval orchestration, permissions, or collaboration should look elsewhere.

Its practical output is normalized Markdown from PDFs, DOCX, PPTX, Excel, EPUB, and community-listed formats such as CSV, RTF, and OpenDocument. Posts also mention a Rust implementation, Node/Python, browser, and CLI bindings, plus an npx example and on-device conversion. These examples indicate plausible integration paths, but they are not a complete official API specification or a compatibility guarantee. Claims such as 4.7 ms, under 5 ms, or 500 files in 1.7 seconds come from social posts or demos and should be treated as illustrative rather than stable benchmarks.