n8n-nodes-anydoc

An n8n community node that converts documents to clean, LLM-friendly GitHub-Flavored Markdown with @firecrawl/anydoc.

Conversion runs locally. It doesn't require credentials or send documents to an external service.

Supported formats

| Family | Extensions | | --- | --- | | Word | .doc, .docx, .docm | | PowerPoint | .ppt, .pps, .pot, .pptx, .pptm, .ppsx, .ppsm | | Excel | .xls, .xlsx, .xlsm, .xlsb | | OpenDocument | .odt, .ods, .odp | | Other | .rtf, .epub, .csv, .pdf |

PDF conversion supports text-based PDFs. Scanned or image-only PDFs require a separate OCR service. Embedded images are represented by their alt text because Markdown can't contain the image bytes.

Installation

This node contains a native dependency and is intended for self-hosted n8n.

Install n8n-nodes-anydoc from Settings → Community Nodes in n8n, or install it in the community nodes directory:

npm install n8n-nodes-anydoc

Restart n8n after installation.

The package requires Node.js 22 or newer. Prebuilt anydoc binaries are available for common Linux glibc and musl, macOS, and Windows environments on x64 and ARM64.

Usage

1. Add a node that produces binary document data, such as an HTTP Request, Form, or Read/Write Files from Disk node.
2. Add the Anydoc node.
3. Set Input Binary Field to the field containing the source document. The usual value is data.
4. Choose whether to return Markdown as text, a .md binary file, or both.
5. Execute the workflow.

Auto detection inspects the document bytes first and then falls back to the source filename and MIME type. CSV has no binary signature, so unnamed CSV input needs either a text/csv MIME type or an explicit Input Format selection.

By default, the node adds:

• Markdown text at json.markdown
• A Markdown file at binary.markdown
• Conversion metadata at json.anydoc

All input JSON and binary fields are preserved. Output items are linked to their corresponding input items.

Limitations

• Plain text, Markdown, HTML, and image files aren't anydoc input formats.
• Scanned PDFs aren't OCRed.
• Embedded image bytes aren't copied into the Markdown output.
• Large documents are loaded into worker memory as a buffer, following n8n's binary-data helper API.
• Because anydoc is an external native dependency, this package isn't eligible for n8n verified community-node distribution under the current no-runtime-dependencies requirement.

Development

npm install
npm test
npm run lint
npm run build

Run a local n8n development instance with:

npm run dev

License

MIT