dokyumi-mcp

MCP (Model Context Protocol) server for Dokyumi document parsing. Gives Claude, ChatGPT, and other AI agents the ability to parse documents.

What This Does

Once installed, your AI assistant can:

• Parse any document — PDFs, images, scans, photos
• Extract structured data — get JSON back with named fields
• Handle any document type — 1099s, W-2s, invoices, receipts, bank statements, medical bills, leases, and more

Setup

Claude Desktop

Add to your claude_desktop_config.json:

{
  "mcpServers": {
    "dokyumi": {
      "command": "npx",
      "args": ["-y", "dokyumi-mcp"],
      "env": {
        "DOKYUMI_API_KEY": "dk_live_your_api_key_here"
      }
    }
  }
}

OpenClaw

mcp:
  dokyumi:
    command: npx
    args: ["-y", "dokyumi-mcp"]
    env:
      DOKYUMI_API_KEY: dk_live_your_api_key_here

Any MCP-Compatible Client

DOKYUMI_API_KEY=dk_live_... npx dokyumi-mcp

Tools Provided

parse_document

Parse a document file and extract structured data.

Parameters:

• file_path (required): Path to the document (PDF, PNG, JPG, TIFF, WebP)
• schema (optional): Specific parsing schema to use

Example conversation:

You: Parse the 1099 form at ~/Documents/tax/1099-nec.pdf

Claude: I'll parse that document for you. [calls parse_document]

The 1099-NEC shows:

• Payer: Acme Corp (EIN: 12-3456789)
• Recipient: John Smith
• Nonemployee compensation: $45,000.00
• Federal tax withheld: $9,000.00

list_schemas

List available parsing schemas in your account.

Get Your API Key

1. Sign up at dokyumi.com
2. Go to Dashboard → API Keys
3. Create a new key

Free tier: 50 extractions/month, no credit card required.

Links

• Dokyumi — Document parsing platform
• CLI Tool — Parse documents from the command line
• API Docs — REST API documentation

License

MIT