Invoice Parser MCP

Parse invoices, receipts, and financial documents into structured JSON — from your AI agent.

Built for the Model Context Protocol. Powered by Claude Vision.

What it does

Extracts structured data from PDF invoices, scanned receipts, and image files. No templates, no OCR configuration — Claude Vision reads the document and returns clean JSON.

Tools

Supported formats

• PDF (invoices, scanned documents)

• PNG, JPG, WEBP (photos of receipts, screenshots)

Authentication

Free tier: 20 parses/month with an API key (get one at plenitudo.ai)

Pay-per-use (x402): No account needed. Send USDC on Base to the wallet address, pass the tx hash as payment_proof.

// x402 payment instructions (returned when no auth provided):
{
  "error": "Payment required",
  "x402": {
    "network": "base",
    "token": "USDC",
    "recipient": "0x9053FeDC90c1BCB4a8Cf708DdB426aB02430d6ad",
    "amount_usdc": 0.05
  }
}

Usage (Claude Desktop / MCP client)

{
  "mcpServers": {
    "invoice-parser": {
      "url": "https://invoice-parser.plenitudo.ai/mcp"
    }
  }
}

Deployment (Railway)

1. Fork this repo

2. Connect to Railway → New Project → Deploy from GitHub

3. Add environment variables:

   • ANTHROPIC_API_KEY — your Anthropic API key

   • INVOICEPARSER_DATA_DIR — /data

   • STRIPE_WEBHOOK_SECRET — from Stripe dashboard

   • STRIPE_PRO_PRICE_ID — from Stripe dashboard

4. Add a persistent volume at /data

5. Deploy

License

MIT