The full upstream README, mirrored here for reference. Install config, tool schemas, adoption signals, and an original overview live on the 0xpdf MCP listing page.
Schema in → PDF in → JSON out for Claude, Cursor, and other MCP clients.
This stdio MCP server wraps the 0xPdf HTTP API (auth, billing, rate limits intact). Agents can parse invoices/forms into a JSON schema, generate schemas, and poll async jobs.

Playground · Docs / MCP setup · Pricing
Get an API key at 0xpdf.io/dashboard.
Or with pip:
| Tool | Purpose |
|---|---|
parse_pdf_sync | Schema-first sync parse |
parse_pdf_stream | Streaming parse (SSE) |
submit_parse_job / wait_for_job_completion | Async jobs |
list_schemas / create_schema | Saved schemas |
generate_schema_from_description | AI schema generation (wallet debit) |
get_pricing_current | Balance / pricing |
health_check | Liveness |
Full list and env vars: see source oxpdf_mcp/server.py and docs.
| Variable | Default | Notes |
|---|---|---|
PDF_PARSING_API_KEY | — | Required for authenticated tools |
PDF_PARSING_API_BASE_URL | https://api.0xpdf.io | Production API |
PDF_PARSING_ALLOWED_API_HOSTS | — | Required in production (api.0xpdf.io) |
PDF_PARSING_REQUIRE_HTTPS | false | Set true for production |
PDF_PARSING_ALLOWED_FILE_ROOTS | CWD | Colon-separated absolute paths for pdf_path |
oxpdf@0xpdf/clientMIT