Ardaro MCP fixture: preserving review flags when an invoice result becomes agent-visible output

[ashevilleislove]

I'm Brent, founder of Ardaro. I am sharing a fixed MCP result that BeeAI tool authors can use when reviewing how an adapter exposes structured evidence to an agent. The interesting case is an invoice with no detected discrepancies that still requires human review.

Connect a Streamable HTTP MCP client to https://agents.getardaro.com/mcp, complete the normal initialization handshake, then make this tool call (it is not a complete client configuration):

{"jsonrpc":"2.0","id":3,"method":"tools/call","params":{"name":"get_invoice_matching_example","arguments":{}}}

This fixed synthetic example is free and needs no wallet. In our direct MCP check on September 13, 2026, the server identified itself as ardaro-agent-services 1.3.0 and listed 18 tools. The fixture returned synthetic_only: true, response.status: "review_required", and:

{
  "outcome": "no_discrepancies",
  "human_review_required": true,
  "delivery_verified": false,
  "payment_approved": false
}

Those four fields are from response.result.summary; they are an excerpt, not the entire tool result. The example's invoice quantity is 2 against an order quantity of 5. Its policy explicitly says earlier invoices and deliveries are not checked. Matching supplied fields therefore does not establish delivery or authorize payment.

For a BeeAI evaluation, inspect the MCP result before and after your tool-output conversion. Keep the review and delivery flags alongside the positive outcome in whatever representation reaches the agent; reducing the result to the string no_discrepancies loses the decision boundary. The fixture also includes comparison rows and source pointers, so a review interface can link a conclusion back to the supplied invoice and purchase-order lines.

This is a reproducible result to evaluate adapter behavior against, not a claim that I have tested a BeeAI integration. Feedback on which fields are most useful to preserve in agent-visible output would help improve the contract.

The live services are read-only and advisory. The separate paid analyses use per-service x402 fees on Base and require owner-authorized payment. This example check made no paid calls. Ardaro does not perform image OCR or post to accounting.

Service documentation

Disclosure: I am affiliated with Ardaro. This post was prepared with AI assistance and checked against the live MCP response. Validation here is direct MCP, not an executed native BeeAI integration.