Put digital PDFs in your agent pipeline

Remote MCP and REST for Cursor, Claude, and backend workflows — pdf_parse, pdf_extract, pdf_split, pdf_ask. Text-layer PDFs only. Not a chat-with-PDF app. Not OCR.

CursorClaude DesktopMCP agents

Built for developers wiring PDFs into agents and pipelines — not office chat tools or scanned-paper OCR.

200 free credits/mo · digital PDFs only · no OCR

mcp.json
{
  "mcpServers": {
    "rosetta": {
      "command": "npx",
      "args": ["-y", "rosetta-mcp"],
      "env": {
        "ROSETTA_API_KEY": "your-api-key",
        "ROSETTA_API_URL": "https://rosettafolio.com"
      }
    }
  }
}

Rosetta API key · env vars ROSETTA_API_KEY and ROSETTA_API_URL

Four endpoints

Method, path, credits — everything you need before opening the docs.

01
POST/api/v2/parse

Parse

Digital PDF → markdown + per-page JSON. Returns docId for downstream calls.

1 credit/page · 0 if cached

02
POST/api/v2/extract

Extract

Structured field extraction from parsed docs using a JSON schema.

2–4 credits/page

03
POST/api/v2/split

Split

AI section detection. Returns page ranges per named section.

2 credits/page

04
POST/api/v2/ask

Ask

Natural-language Q&A across one or more parsed documents.

3 credits/page total

Try before you integrate

Upload a digital PDF — parse, extract, or split in the browser. No API key required.

Document

Drop PDF here

or open full playground

→ markdown output

→ per-page JSON contents

→ docId for extract / split / ask

response preview

{

"success": true,

"docId": "doc_abc123",

"pageCount": 12,

"markdown": "## Page 1\n..."

}

Ship in three steps

  1. 01

    Get an API key

    Sign up — 200 free credits/month, no card.

    Sign up
  2. 02

    Parse a PDF

    POST /api/v2/parse with a URL or file upload — or try pdf to markdown in the browser first.

    pdf to markdown
  3. 03

    Use the docId

    Pass to extract, split, or ask endpoints — or install MCP for agents.

    Install MCP

Cursor & Claude, out of the box

pdf_parse, pdf_extract, pdf_split, pdf_ask — four MCP tools, one npx install. No REST wrapper for your agent.

MCP setup
mcp.json
{
  "mcpServers": {
    "rosetta": {
      "command": "npx",
      "args": ["-y", "rosetta-mcp"],
      "env": {
        "ROSETTA_API_KEY": "your-api-key",
        "ROSETTA_API_URL": "https://rosettafolio.com"
      }
    }
  }
}

pdf to markdown, then the API

The free browser converter is pdf to markdown— one file, 5 pages, no sign-in. Same parse as REST and the PDF MCP server. 200 free credits/month after you create an API key, then paid plans. Related:

Credits buy API operations

One credit ≈ one page processed on most endpoints. Cached parses are free. All four endpoints included on every plan.

Parse1 credit / page
Extract (≤5 fields)2 credits / page
Split2 credits / page
Ask3 credits / page total
Cached re-parse (same docId)0 credits
Full pricing

Free

$0

200/mo

Starter

$49

3,000/mo

Pro

$99

10,000/mo

Scale

$249

30,000/mo