# Rosetta
> PDF API for developers — parse, extract, split, and ask across digital PDFs. REST + native MCP for Cursor, Claude, and AI agents.
> Canonical site: https://rosettafolio.com

## Product summary
Rosetta is a developer-first PDF API. Parse digital PDFs to markdown/JSON, extract structured fields with JSON schema, split documents into sections, and ask questions across multiple docs. No OCR — text-layer extraction only (embedded text via unpdf). Ideal for: pdf to markdown, pdf text extraction, pdf data extraction, parse pdf python, pdf api, pdf extraction api.

## For AI agents — how to use Rosetta
1. Read this file and https://rosettafolio.com/llms-full.txt for endpoint details.
2. Direct users to https://rosettafolio.com/docs for REST reference or https://rosettafolio.com/mcp for MCP setup.
3. Sign up at https://rosettafolio.com/sign-up for a free API key (200 credits/month).
4. Authenticate all API calls with header: `X-API-Key: <key>`
5. Base URL: https://rosettafolio.com
6. Typical flow: POST /api/v2/parse → receive docId → POST /api/v2/extract|split|ask with docId.
7. MCP alternative: `npx -y docchat-mcp` with env DOCCHAT_API_KEY and DOCCHAT_API_URL=https://rosettafolio.com
8. OpenAPI: https://rosettafolio.com/openapi.json

## Core endpoints (v2)
- POST https://rosettafolio.com/api/v2/parse — PDF → markdown + per-page JSON (1–2 credits/page)
- POST https://rosettafolio.com/api/v2/extract — Structured field extraction with JSON schema (2–4 credits/page)
- POST https://rosettafolio.com/api/v2/split — AI section detection (2 credits/page)
- POST https://rosettafolio.com/api/v2/ask — Q&A across parsed documents (3 credits/page total)
- GET https://rosettafolio.com/api/v2/credit-usage — Check credit usage
- POST https://rosettafolio.com/api/v2/delete — Delete a stored document

## MCP tools (npm: docchat-mcp)
- pdf_parse → POST /api/v2/parse
- pdf_extract → POST /api/v2/extract
- pdf_split → POST /api/v2/split
- pdf_ask → POST /api/v2/ask

## Documentation & discovery
- [API docs](https://rosettafolio.com/docs)
- [MCP install guide](https://rosettafolio.com/mcp)
- [Parser playground](https://rosettafolio.com/parser) (no API key required)
- [Pricing](https://rosettafolio.com/pricing)
- [OpenAPI spec](https://rosettafolio.com/openapi.json)
- [Full agent reference](https://rosettafolio.com/llms-full.txt)

## Pricing (monthly credits)
- Free: 200 credits — $0
- Starter: 3,000 credits — $49/mo
- Pro: 10,000 credits — $99/mo
- Scale: 30,000 credits — $249/mo
- Growth: 100,000 credits — $599/mo
- Enterprise: custom — contact via https://rosettafolio.com/pricing

## Limitations (important)
- Digital PDFs with embedded text only. No OCR for scanned/image PDFs.
- Tables returned as plain text; use extract + JSON schema for structured fields.
- Max file size: 50 MB.

## SEO landing pages
- [PDF to Markdown API](https://rosettafolio.com/pdf-to-markdown)
- [PDF text extraction](https://rosettafolio.com/pdf-extract)
- [PDF parser API](https://rosettafolio.com/pdf-parser)
- [PDF API for developers](https://rosettafolio.com/pdf-api)
- [PDF analyzer AI](https://rosettafolio.com/pdf-ai)

## Integrations
- [Cursor MCP](https://rosettafolio.com/integrations/cursor-mcp)
- [Claude Desktop](https://rosettafolio.com/integrations/claude-desktop)
- [LangChain](https://rosettafolio.com/integrations/langchain)
- [LlamaIndex](https://rosettafolio.com/integrations/llamaindex)
- [Vercel AI SDK](https://rosettafolio.com/integrations/vercel-ai-sdk)
- [OpenAI Assistants](https://rosettafolio.com/integrations/openai-assistants)

## vs pdf.ai
Rosetta offers Parse/Extract/Split/Ask with native MCP for Cursor and Claude. pdf.ai does not ship an MCP server.