Contract Parser

What Document to JSON – PDF Invoice/Statement/Contract Parser MCP server does

Document to JSON – PDF Invoice/Statement/Contract Parser MCP server accepts a public PDF URL and returns extracted data as JSON. It is designed for common business documents, including invoices, bank statements, and contracts, while a generic mode handles full text and tables from other document types.

Invoice results can include seller and buyer details, invoice numbers, dates, line items, totals, VAT, IBANs, and payment information. Bank statement extraction covers transactions, balances, fees, and account-holder details. Contract extraction can identify parties, dates, financial terms, jurisdiction, and key clauses with summaries. Results can also include a document type and confidence value.

How it works

The MCP tools send a PDF URL and document selection to an Apify-hosted actor. The actor uses AI-based document understanding rather than fixed templates or regular expressions, allowing it to process different layouts and vendors. OCR is included for scanned PDFs.

Available OCR languages are English (eng), Italian (ita), and Spanish (spa). Multiple languages can be combined, such as eng+ita+spa, which is the documented default. Sample extractions in the README take approximately 15–20 seconds, although processing time may vary.

The service also supports validation options for extracted content. For example, the REST example enables validate_totals for an invoice. Pricing is described as pay per use through Apify: invoices, bank statements, and contracts have separate per-document prices, while generic parsing is free during the stated launch period.

Setup and configuration

Provide a publicly accessible PDF URL, choose the document type, and run the actor. The MCP endpoint is hosted at https://mcp.apify.com/?actors=opportunity-biz/document-to-json-mcp. Claude Code can add it using an HTTP transport. Claude Desktop and Cursor can connect through mcp-remote and an Apify authorization header.

An Apify token is used in the documented Claude Desktop, Cursor, REST, Python, and JavaScript examples. No local PDF path is described; the documented input is a public file_url.

Tools and capabilities

Document to JSON – PDF Invoice/Statement/Contract Parser exposes these MCP tools:

• parse_invoice for invoice fields, line items, tax, totals, and payment details.
• parse_bank_statement for transactions, balances, fees, and account information.
• parse_contract for parties, dates, clauses, financial terms, and jurisdiction.
• parse_generic_document for general document text and tables.

The same actor can also be called through its REST API or Apify client libraries. Extracted data is made available in an Apify dataset, and the REST endpoint can return the result directly with run-sync-get-dataset-items.

Limitations and notes

The documented workflow requires a public PDF URL. OCR language support is limited to English, Italian, and Spanish as documented. Extraction prices apply by document type, and Apify access requires a token in the integration examples.

These privacy statements describe the hosted actor workflow and should be evaluated against the deployment and data-handling requirements of your application.