Pickanydoc: Convert 14 Office Formats to Markdown in 5ms

A Rust-based library that converts 14 document formats, including Word, PowerPoint, Excel, and PDF, into GitHub-Flavored Markdown. Developed by Firecrawl, it replaces the document preprocessing step before feeding into LLM pipelines with a single tool. It is designed to ensure all input formats produce Markdown with the same structure, eliminating output variations based on source file type.

The conversion speed has a median of 4.4ms per document, making it more than 10 times faster than competing tools. It runs in pure Rust without dependencies on ML models or external services, and PDF text extraction is also handled locally. It precisely preserves document structures such as merged table cells, footnotes, formulas (converted to LaTeX), and embedded images, and supports content-based format detection rather than relying on file extensions.

It provides four bindings: Node.js, Python, Rust, and WebAssembly. In browser environments, it runs via WebAssembly, processing files locally without sending them to a server. It is also offered as an Agent Skill, integrating with AI agents like Claude Code and Cursor to allow them to read document files directly.

It is suitable for data pipelines receiving mixed office documents that require consistent Markdown output. There is no need to combine different conversion tools for each format or call slow external APIs as in traditional approaches. It is released under the MIT license and is an active project with 18.2k stars.