# PayeeProof Agent Skill

Use PayeeProof when an agent is about to make an autonomous machine payment and needs to verify that the endpoint, claimed service, and public payment destination agree.

## Workflow

1. Discover the target service and its public payment metadata.
2. Obtain or observe its HTTP 402 payment challenge. For POST, PUT, PATCH, or DELETE targets, observe the target challenge yourself; PayeeProof will not actively send those methods.
3. Call `payeeproof.check_payee` or POST `/v1/payee/check`.
4. If the request is eligible, pay PayeeProof exactly $0.01 over a compatible stablecoin rail. MCP uses MPP/Tempo; HTTP can also use x402 when configured.
5. Read `assessment`, individual checks, provenance, history, limitations, and `deterministicDigest`.
6. Proceed only according to the calling agent's own spending and merchant policy.

## Safety boundary

Never send the target merchant's payment credential, Authorization header, cookies, private keys, card data, or arbitrary request body to PayeeProof. PayeeProof does not authorize spending. A `consistent` result is not a delivery guarantee, reputation score, KYC/AML/sanctions result, legal identity verification, fraud certification, or legal advice. Handle `review_required`, `conflict`, and unknown checks according to caller policy.