convert-documents-to-markdown Skill: Read Word, Excel, PowerPoint and PDF in Claude
Lets Claude read Word, Excel, PowerPoint, OpenDocument, RTF, EPUB, CSV and PDF files by converting them to Markdown with the anydoc tool.
- Installs
- 12.9K
- on skills.sh
- Stars
- 23.3K
- on GitHub
- Updated
- Oct 2026
- last repo commit
convert-documents-to-markdown is the agent skill that ships with Firecrawl's anydoc tool. When a task needs the contents of an office document, spreadsheet, presentation, ebook or PDF that the agent cannot read directly, it runs npx -y @firecrawl/anydoc on the file and gets GitHub-Flavored Markdown back, either on screen or in a file.
The skill needs Node 20 or newer. Text-based documents convert on your machine. Scanned PDFs need OCR, which anydoc does not do, so the tool stops with exit code 3. Only if Claude reruns it with --ocr hosted does the whole document go to Firecrawl Parse, a hosted service.
Good for
- Developers whose agent keeps hitting .docx, .xlsx or .pptx files it cannot open
- Analysts who want a spreadsheet or slide deck turned into text a model can read
- Researchers feeding PDFs and ebooks to Claude as context
- Teams reading documents with sensitive content who need to know when a file leaves the machine
Install convert-documents-to-markdown
Claude Code
- 1
Copy this prompt.
Install the convert-documents-to-markdown skill in Claude Code. Run this command: npx skills add firecrawl/anydoc Then check that it loaded and tell me how to use it.
- 2
Paste it into Claude Code. It shows the change and asks before running it.
Requires Node.js 20+.
The skill runs the anydoc command-line tool through npx, so it needs Node.js 20 or newer on the machine where the agent runs. That is why there is no Claude app upload route here.
From the project README. Source.
What's in the skill
When it loads
Convert Word (.doc, .docx), PowerPoint (.ppt, .pptx), Excel (.xls, .xlsx), OpenDocument (.odt, .ods, .odp), RTF, EPUB, CSV, and PDF files to GitHub-Flavored Markdown. Use when a task needs the contents of an office document, spreadsheet, presentation, ebook, or PDF you cannot read directly.
Key facts
- Author
- Firecrawl
- Repository
- firecrawl/anydoc
- Installs on skills.sh
- 12,907 as of 2026-11-02
- GitHub stars
- 23,291 for the whole repo, 2026-11-02
- License
- MIT (repo metadata)
- Last commit to the repo
- 2026-10-23
- Files
- 1 file (SKILL.md)
- Author source
- The SKILL.md frontmatter lists firecrawl as author
What it includes
npx -y @firecrawl/anydoc <file> prints Markdown
Runs the anydoc tool on a file and prints GitHub-Flavored Markdown. The format is detected from the file content.
Try: “Read budget.xlsx and summarize the Q3 sheet.”
Runs code ·
Convert to stdoutnpx -y @firecrawl/anydoc <file> -o out.md saves the Markdown
Recommended for large documents: write to a file, then read only the parts needed instead of streaming everything into context.
Runs code ·
Write to a file (-o)Reads from stdin when the format cannot be detected, such as CSV
Pass --format <name> only for CSV from stdin, or when the extension is missing or wrong.
Runs code ·
Read stdin (--format)0 success, 1 convert failed, 2 usage error, 3 pages need OCR
Failures print one anydoc: message line to stderr, and the tool never prompts.
Exit codesSends a scanned PDF to Firecrawl Parse for OCR
Anydoc does no OCR itself. With --ocr hosted the document is sent to Firecrawl Parse, which needs no signup. A --api-key or FIRECRAWL_API_KEY gives higher limits.
Network ·
Hosted OCR (--ocr hosted)In Node, Python or Rust code, use the library instead of the CLI
SKILL.md points to @firecrawl/anydoc on npm, firecrawl-anydoc on PyPI and anydoc on crates.io, all with the same to_markdown API.
Library hint
From the official docs
Permissions
Medium risk
Text documents convert locally, but a scanned PDF can be sent whole to Firecrawl Parse when the OCR option is used. Keep confidential scans out of that option.
What it can do
- Read your data
- Create and change things
- Delete things
- Read-only mode
Watch out for
Scanned PDFs can leave your machine
With --ocr hosted the whole document goes to Firecrawl Parse. Do not allow it for confidential files. Without that flag, the README says conversion runs locally. github.com
It runs a package fetched by npx
SKILL.md runs npx -y, which installs the package without asking. Pin a version or install it globally if you want control. github.com
Optional API key
FIRECRAWL_API_KEY, if set in your environment, is used for hosted OCR limits. Leave it unset unless you want that. github.com
Technical details
| Talks to | Firecrawl Parse (firecrawl.dev), only when run with --ocr hosted |
|---|---|
| Scopes |
|
Questions
What does the convert-documents-to-markdown skill do?
It teaches the agent to turn Word, PowerPoint, Excel, OpenDocument, RTF, EPUB, CSV and PDF files into Markdown by running the anydoc command-line tool.
How do I install it in Claude Code or Codex?
Use the install box at the top of this page. It needs Node.js 20 or newer. The Claude app upload route is left out because the skill runs a command-line tool through npx.
Does it need an API key?
No. FIRECRAWL_API_KEY is optional and only raises the limits of the hosted OCR option, which Claude uses only if it runs the tool with --ocr hosted.
Do my documents leave my machine?
Text-based documents are converted locally, according to the anydoc README. A scanned PDF is sent in full to Firecrawl Parse only if the tool is run with --ocr hosted.
Is it safe?
The skill runs npx -y @firecrawl/anydoc, which downloads and runs the package on first use, and it can write a Markdown file with -o. Decide whether you are comfortable with that package and with the hosted OCR option before allowing it.
Can it read scanned PDFs?
Not on its own. The tool exits with code 3 for scanned or image-only pages. Re-running with --ocr hosted sends the document to Firecrawl Parse.
Information
updated Nov 2, 2026
- Repository
- firecrawl/anydoc
- Folder
skills/convert-documents-to-markdown
Sources
Docs, issue trackers and discussions this page is built from.
Show all 2Hide
- anydoc README · github.com
- SKILL.md · github.com