🚨 Firecrawl Just Open-Sourced a Massive Utility for AI Devs
Source: Charly Wargnier on LinkedIn (11 comments)
The Tool: anydoc
A Rust library that converts almost any office document into clean Markdown in milliseconds.
- No models involved — pure speed, no LLM token cost
- Supports Word, PPT, Excel, PDF, EPUB, CSV
- Uses one shared document model for consistent output
- Runs locally via WebAssembly or CLI
- MIT licensed, ~10.8k stars
- Can be dropped straight into Claude Code or Cursor as an agent skill with a single command
- 100% free and open-source
Why It Matters
- Massive token savings for LLM/RAG pipelines: document → Markdown → LLM means no formatting/metadata noise in context
- Local + WASM = private, fast, no API costs
- Direct competitor/complement to Microsoft’s MarkItDown (commenters noted differences — anydoc is pure Rust speed, no models)
Repo
Firecrawl’s anydoc on GitHub (link via lnkd.in/e5K22xP6).
Comments Snapshot
- MarkItDown comparison: “how does that differ from microsoft/markitdown?”
- Skeptic: “struggles converting PDF to markdown for simple research papers”
- Others planning to integrate it into their projects
Related
- linkedin-charlywargnier-10-stage-ai-mastery — same author’s 10-stage AI mastery roadmap
- Microsoft MarkItDown — similar tool, Python-based