docimprint

MCP document server for AI agents

What MCP tools does DocImprint expose?

DocImprint MCP tools include extract_text, extract_tables, parse_invoice, check_claims, extract_structured, verify_bundle, search_collection, and ask_collection — all with citation-grounded document intelligence.

How do I connect Claude Desktop to DocImprint MCP?

Use the mcp-remote stdio bridge with Authorization: Bearer dr_live_… pointing to https://api.docimprint.com/mcp. See /docs/agents for full configuration examples.

Does MCP support verifiable evidence bundles?

Yes. MCP extract tools return the same agent envelope as REST — bundle_id, manifest SHA-256, and optional Merkle citation proofs. verify_bundle confirms integrity offline-capable via GET /v1/keys.

Model Context Protocol

Plug into Claude, Cursor, CrewAI, LangGraph — zero glue code.

Native MCP server exposes document tools directly to any compatible agent runtime.

jsonClaude Desktop (mcp-remote)
{
  "mcpServers": {
    "docimprint": {
      "command": "npx",
      "args": [
        "-y",
        "mcp-remote",
        "https://api.docimprint.com/mcp",
        "--header",
        "Authorization:${DOCIMPRINT_AUTH}"
      ],
      "env": {
        "DOCIMPRINT_AUTH": "Bearer dr_live_YOUR_KEY"
      }
    }
  }
}

Requires Node 18+ and DOCIMPRINT_AUTH with your API key. MCP tools use the same intelligence layer as REST; discovery and verification endpoints remain free.

  • extract_textPDF or image → raw text via OCR
  • extract_tablesPDF or image → Markdown tables
  • parse_invoiceReceipt or invoice → structured JSON (merchant, date, line_items, total)
  • check_claimsDocument + claims[] → supported / contradicted / not_found with citations
  • extract_structuredDocument + JSON schema → typed fields with per-field citations
  • summarize_documentDocument → summary + key_points[]
  • verify_bundlebundle_id → integrity check (hash match / tampered / not found)
  • extract_urlHTTPS URL → page text
  • summarize_urlHTTPS URL → summary + key_points[]
  • qa_urlHTTPS URL + question → cited answer
  • translate_urlHTTPS URL + target language → translated text
  • create_collectionCreate a named document corpus for search and Q&A
  • search_collectioncollection_id + query → ranked semantic search results
  • ask_collectioncollection_id + question → cited answer across all documents

Related