Overview
An AI agent discovers your store, pays you in USDC over the x402 standard, and you deliver. Veto Checkout handles the payment and checks every order against your rules before a cent reaches you.
USDC is a dollar-pegged stablecoin and x402 is the open standard agents use to pay over the web – so you get paid in plain dollars, and the agent's tools handle the crypto part.
Before any money reaches you, Veto confirms the agent is allowed to spend and that the order fits the rules you set – so you only accept payments you'd want to keep. Anything that doesn't fit never settles, and both sides walk away with a signed receipt you can use as dispute evidence.
Veto governs · the rail executes.
Mandate verification and policy run locally – zero latency, and they work even if the hosted services are down. The rail only ever executes a spend the gate already approved.
What you get
Open to all agents
A presented mandate is optional and polymorphic (veto | ap2 | acp | none). A
Veto-governed agent is simply the highest-trust input – never a requirement.
Non-custodial
Veto never holds funds. You bring your own receiving address; the only key Veto holds signs receipts, not value.
Rail-agnostic, x402 first
The Rail is a pluggable interface. v1 ships x402 (HTTP-402 + stablecoin) and mock
(offline tests); card is a typed stub.
The acceptance gate
Verify the mandate, check replay, look up reputation, run policy – all before any money is captured. The order is the chargeback defense.
Start here
Quickstart
A fully self-describing agentic checkout in 30 seconds, then settle a mock order.
Agentic Checkout Protocol
Every endpoint, status code, and reason code – the self-describing wire protocol agents talk to.
SDK · CLI · MCP
The @veto-protocol/checkout SDK, the veto CLI, and the MCP server.
Reason codes
The stable, machine-actionable rejection vocabulary an agent keys off.