anydoc
anydoc converts Word, PowerPoint, Excel, OpenDocument, RTF, EPUB, CSV, and PDF into clean GitHub-Flavored Markdown. Built in Rust with Node, Python, and WebAssembly bindings, it delivers consistent, asset-aware output and millisecond latency for agents, RAG pipelines, and developer workflows.
Overview
Drop office files into anydoc and receive structured, GFM-clean Markdown with headings, lists, tables, footnotes, links, and asset references intact. The library detects formats from bytes, normalizes structure through a shared model, and serializes once for predictable, identical output.
Consistent Markdown from Every Office Format
Ideal for engineers building RAG and retrieval pipelines, data ingestion and ETL jobs, agents that must read arbitrary documents, and developer tools that need Markdown as a stable interchange. Teams replacing brittle, format-specific scripts or heavyweight headless-office stacks benefit from a single, predictable output. Back-end services process documents off-thread; Python data workflows keep the GIL free; browser apps convert locally with WASM for privacy. If your requirement is clean, structured Markdown with minimal surprises, anydoc is purpose-built.
- Parses multiple office formats into one shared, consistent document model.
- Serializes once to GitHub-Flavored Markdown with identical behavior across inputs.
- Detects format from file bytes to handle mislabeled or extensionless documents.
- Exposes Node, Python, and WebAssembly bindings for flexible deployment.
- Preserves embedded assets and metadata alongside the rendered Markdown.
Why Teams Choose anydoc
Engineering Notes and Guarantees
Use the CLI with npx @firecrawl/anydoc to convert files or pipe stdin. For Node.js, install the package and call toMarkdown or toDocument; Python uses pip install firecrawl-anydoc with to_markdown or to_document. The WebAssembly build runs entirely in-browser after initialization. anydoc also ships as an Agent Skill; add it with npx skills add firecrawl/anydoc so agents seamlessly convert encountered documents. Format detection works from bytes, while CSV accepts an explicit format when signatures are absent. Error variants distinguish unsupported, encrypted, malformed, or resource-limited inputs for robust pipeline handling.
One serializer, one output: predictable, GFM-clean Markdown no matter which document went in.
Getting Started and Integration
anydoc replaces glue scripts and heavyweight office stacks with a compact, local, format-spanning converter. The shared model plus single Markdown serializer ensures consistency; Node, Python, and WASM bindings meet teams where they run. Fast, deterministic behavior, asset handling, and explicit error variants make it easy to operate at scale without surprises.
Open the tool and review its core product experience.
Create your account or access your existing workspace.
Use your own task to judge speed, quality, and fit.
Check similar AI tools before making a final decision.
anydoc
Firecrawl builds high-performance document parsing and developer tooling for AI pipelines. The team focuses on deterministic, low-latency extraction in Rust with production-ready bindings, powering both open-source libraries and hosted services used in agents and RAG systems.
anydoc is a fast Rust library that turns heterogeneous office documents into consistent, GitHub-Flavored Markdown. Every supported format—DOC/DOCX, PPT/PPTX, XLS/XLSX, ODT/ODS/ODP, RTF, EPUB, CSV, and text-based PDF—parses into a shared document model before a single serializer renders Markdown, so lists, tables, anchors, and footnotes behave identically across sources. Content-based format detection reads file bytes, not extensions, reducing ingestion errors in pipelines. Bindings for Node.js and Python keep runtimes responsive (libuv thread pool, GIL release), while a WebAssembly build runs fully client-side in the browser. Embedded assets are preserved on the document model and emitted as appropriate Markdown references. Designed for throughput, typical conversions complete in single-digit milliseconds without external services or ML. It powers Firecrawl’s own parsing stack and slots cleanly into agents, RAG, and ETL jobs that require predictable, structured text.
Visit anydoc
Comments (0)
No Comments Found