Rosetta

PDF API for Developers

PDF MCP + REST for agent builders — parse to markdown, extract fields, split, ask. Remote MCP or REST for Cursor, Claude, and pipelines. Digital PDFs only. Not chat-with-PDF. Not OCR.

Four REST endpoints and matching MCP tools: parse, extract, split, ask. Parse is pdf to markdown plus per-page JSON. Start with 200 free credits/month (no card), then paid plans. Browser preview lives on the pdf to markdown page. Digital text-layer PDFs only.

Digital PDFs with embedded text · no OCR · uploads up to 4.5 MB

Install MCP server

Parse API

PDF to markdown and per-page JSON from digital PDFs

pdf to markdown →

MCP server

MCP tools pdf_parse, pdf_extract, pdf_split, pdf_ask in Cursor and Claude

Install MCP server →

200 free credits

API key is free. 200 credits/month, then paid plans — not a free trial

API docs →

FAQ

Does Rosetta support OCR for scanned PDFs?
No. Rosetta extracts text from digital PDFs with an embedded text layer only. Scanned image PDFs are not supported.
How do I parse a PDF in Python?
POST to /api/v2/parse with requests or httpx, passing your PDF file or URL and X-API-Key header. See docs for a full Python example.
Is there an MCP server for Cursor?
Yes. Install via npx -y rosetta-mcp (npm 0.2.2+) with ROSETTA_API_KEY and ROSETTA_API_URL. Tools: pdf_parse, pdf_extract, pdf_split, pdf_ask, pdf_credit_usage, pdf_delete. Local stdio also accepts a file path.
Can I try pdf to markdown before signing up?
Yes. Use the free pdf to markdown tool — one conversion, up to 5 pages, max 4.5 MB upload. Then 200 free API credits/month. Digital PDFs only. Not a free trial.