n8n-nodes-anydoc

An n8n community node that converts Word, PowerPoint, Excel, PDF, RTF, EPUB, CSV, and OpenDocument files into clean GitHub-Flavored Markdown, powered by @firecrawl/anydoc.

Perfect for feeding office documents into LLM/AI Agent nodes without dealing with format-specific parsers.

n8n is a fair-code licensed workflow automation platform.

Table of contents

• Installation
• Operations
• Node inputs
• Node outputs
• Supported formats
• Coolify / self-hosted Docker notes
• Troubleshooting
• Development
• License

Installation

Follow the official community-nodes install guide, then enter:

n8n-nodes-anydoc

as the npm package name in Settings → Community Nodes → Install.

The node will appear in the palette as AnyDoc.

Operations

• Convert to Markdown — reads a document from a binary property and returns Markdown as a JSON field.

Node inputs

| Field | Default | Description | |---|---|---| | Input Binary Field | data | Name of the binary property holding the source document (typical output of Read Binary File, HTTP Request, Google Drive, etc.). | | Output Field | markdown | Name of the JSON field the resulting Markdown will be written to. | | Options → Format Hint | Auto-Detect | Override auto-detection. Required for CSV (which has no content signature). | | Options → Include Source Metadata | false | Adds sourceFileName and detectedFormat to the output JSON. | | Options → Keep Input JSON | true | Merge the incoming item's JSON fields into the output. Disable to emit only the markdown field. |

Node outputs

A single main output. Each output item's JSON contains at least:

{
	"markdown": "# Report Q4 …\n\n| … | … |\n| --- | --- |\n…"
}

When Include Source Metadata is on:

{
	"markdown": "…",
	"sourceFileName": "report.docx",
	"detectedFormat": "docx"
}

If the node runs with Continue On Fail enabled and conversion fails (encrypted file, unsupported format, image-only PDF, …), the item becomes:

{ "error": "…", "code": "Encrypted" }

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 | | Rich Text Format | .rtf | | EPUB | .epub | | CSV | .csv (needs Format Hint = CSV) | | PDF | .pdf (text-based only — no OCR) |

Image-only / scanned PDFs return Unsupported — use Firecrawl Parse or a dedicated OCR service for those.

Coolify / self-hosted Docker notes

This node depends on @firecrawl/anydoc, which ships prebuilt native binaries (Rust via napi-rs). It needs to match your container's architecture.

• Official n8nio/n8n images run on Debian (glibc) for both linux/amd64 and linux/arm64 — the anydoc prebuild resolves automatically. No extra steps needed.
• Make sure /home/node/.n8n is a persistent volume in Coolify (default in n8n's Coolify template). The community package is installed there, so a rebuild without a volume will wipe it.
• If you self-hosted on Alpine, the glibc prebuild will not run — switch to the Debian image or bake anydoc into a custom image with musl support.

Example minimal docker-compose.yml snippet Coolify can consume:

services:
	n8n:
		image: n8nio/n8n:latest
		environment:
			- N8N_COMMUNITY_PACKAGES_ENABLED=true
			- N8N_COMMUNITY_PACKAGES_ALLOW_TOOL_USAGE=true
		volumes:
			- n8n_data:/home/node/.n8n
volumes:
	n8n_data:

Troubleshooting

| Symptom | Cause / fix | |---|---| | Cannot find module '@firecrawl/anydoc-linux-x64-gnu' | The container is Alpine (musl). Use the Debian n8n image (n8nio/n8n:latest), which is glibc. | | Error code Encrypted | The document is password protected. anydoc refuses to guess. Decrypt upstream. | | Error code Unsupported on a PDF | Image-only PDF. Use OCR (Firecrawl Parse, Tesseract, etc.). | | Empty markdown from CSV | Set Options → Format Hint = CSV. CSV has no content signature to auto-detect. | | Tables come out weirdly aligned | anydoc emits GFM tables — Markdown viewers with strict column widths look strange; the rendered output is fine. |

Development

git clone https://github.com/your-github/n8n-nodes-anydoc
cd n8n-nodes-anydoc
npm install
npm run build

To test in a local n8n:

npm link
cd ~/.n8n/nodes    # or wherever custom nodes live
npm link n8n-nodes-anydoc
n8n start

Publish:

npm login
npm publish --access public

The package will then be installable through n8n's Community Nodes UI.

License

MIT — matches the upstream @firecrawl/anydoc license.