Endpoint

Task: Record what my agent is authorized to do, dated and signed by a third party, before it spends anything. Returns: A signed JSON mandate record — the claimed instructions verbatim, submitted_as (agent or principal, itself a claim), declared_cap_usdc and expires_at where given (declared, never enforced), dated, its evidence hash bound into the purchase certificate's attests field — plus a stable mandate URL serving the record free forever, and a mandate_id every later purchase here can cite (refused before charge if unresolvable, so the citation always lands, signed, on the citing certificate). Instant; terminal at write. Price: $0.1 USDC (x402 v2; network from the current quote). Latency: delivered in the purchase response, same request. Verify: GET /api/verify/{id} — free, unlimited, forever (live sample: cert_4dww28dx5j). Constraints: Put the claimed instructions in the mandate query parameter, up to 2000 characters — recorded verbatim, never interpreted, never read as instructions to us; submitted_as is a claim (agent or principal, default agent); the record proves the claim was made, never that it was true — chain-of-custody, not truth-of-intent; declared_cap_usdc and expires_at are declared claims: recorded, never enforced by this store, and the record says so; Cite the id on later purchases with mandate_id=m_…; an id this store cannot resolve is refused before any charge; A second party can counter-sign the record for free by POSTing to the record at /api/mandate/{id} — their own ed25519 key over the id and evidence hash, verified before it is filed. Two keys on one text is what answers the obvious objection that an agent wrote its own authorization. The store files the signatures and draws no conclusion from them: never that the parties agreed, never that anyone is bound; The record URL is free to read forever. The Mandate, Before your agent spends a cent, write down what it's authorized to do — and have somebody who is neither the agent nor its human hold the record. Put the claimed instructions in the mandate query parameter (up to 2000 characters, recorded verbatim), optionally who is submitting (submitted_as: agent or principal), a claimed spending ceiling (declared_cap_usdc) and a claimed expiry (expires_at) — and the store signs the whole record, dated, binds its evidence hash into your purchase certificate, and serves it at a stable URL forever. Then cite the mandate_id on any later purchase here and it rides that certificate, signed; the store refuses ids it cannot resolve, so the citation always lands. Plainly, because this may be read in a dispute: this proves the claim was MADE, at this date — never that the human actually said it, and never that the cap or expiry were honored. Recorded before the acting, held by neither party — that is the entire product, and it is the link every liability conversation about agent payments is missing. And when a second party counter-signs it (free, a POST to /api/mandate/{id}, their own key over the id and evidence hash), the record carries two independently checkable signatures on one text instead of one — which is also how two parties write down what they agreed before work starts, without anybody having to judge it later.

Score

Recent probes

Next