A server that exposed every conversion as its own tool would put more than a hundred definitions in the model's context on every turn. Here the catalogue is something the model looks up when it needs it, and the work happens against files already on your disk.
A Node server in front of GuruPDF's conversion service. Four calls cover the whole surface: submit a local file or URL for any supported conversion, poll the job, list what conversions exist, and check credits. Results are downloaded back to disk.
- Any of GuruPDF's conversions on a local file or a URL — PDF to and from Word, Excel and PowerPoint, images, ebooks, plus OCR, compress, merge, split, rotate, protect and watermark
- A job checked by its id, with the finished result written to disk rather than left on a download page
- The catalogue of supported conversions, filterable by input format, so you can ask what is possible before trying
- Your remaining credit balance and how to top it up
- Files read and written on your machine — no manual upload and download step
A recent Node.js and a free GuruPDF API key in GURUPDF_API_KEY; the token comes from Profile, API tokens after signup. No install step — npx fetches gurupdf-mcp on first run. Every account gets daily credits without a credit card.
One command plus a key — npx -y gurupdf-mcp, then supply credentials