Home / Use Cases / Catch a fraudulent invoice before paying it
Use case3-step workflow

Catch a fraudulent invoice before paying it

Pull structured data out of an invoice, check it for fraud signals, and match it against purchase orders.

Workflow

Step 1 of 3

Invoice Extract

Extract line items, totals, and vendor details from raw invoice text.

invoice_extract({"text": "hello from an agent"})
Step 2 of 3

Invoice Fraud Check

Flag anomalies against previously seen invoices from the same vendor.

invoice_fraud_check({"invoice": {}})
Step 3 of 3

Invoice Match

Match the invoice against open purchase orders before approving payment.

invoice_match({"invoice": {}})

Install

One command wires up every tool below for Claude Code:

claude mcp add --transport http wingman https://wingmanprotocol.com/mcp

Claude Desktop, Cursor and Continue configs are on /install. A free key (500 calls/month, no card) is on /pricing; beyond that it is x402 pay-per-call in USDC on Base.