Budget authority
Daily budgets, max-per-call rules, provider allowlists, and blocked actions.
AI Commerce Guardrails
Give your AI freedom without unlimited authority.
IntentLock turns vague user intent into enforceable agent mandates before money moves. Agents ask IntentLock for permission before they spend, hire providers, call paid MCPs, or release escrow.
IntentLock / payment gateX Layer · eip155:196
Approved
{
"decision": "approved",
"reason": "Provider is allowlisted, request is within max-per-call, and budget remains.",
"remainingBudget": "29.70 USDt",
"receiptRequired": true
}Blocked
{
"decision": "blocked",
"reason": "Payment exceeds max-per-call and provider is not allowed by the mandate.",
"requiredAction": "request_user_approval"
}The trust layer for AI commerce.
Every commercial agent workflow needs a permission layer. IntentLock defines what the agent can spend, who it can pay, what evidence it must collect, and when delivery is good enough for settlement.
Acceptance criteria
Clear delivery requirements before another ASP is hired or escrow is released.
Evidence trail
Receipts, action logs, provider responses, and dispute packets for review.
Live Policy Proof
Watch IntentLock make real decisions.
This button calls a live API route that uses the same mandate engine behind the A2MCP tools. It creates a mandate, approves an allowed payment, blocks an unsafe action, scores a deliverable, and returns receipts.
IntentLock / live engineReady
MandateWaiting for proof runBudget -- | Max per call --
Safe paymentPendingApproved result appears here.
Unsafe actionPendingBlocked result appears here.
DeliverablePendingDelivery review appears here.
A2MCP tools
create_mandateDefine rules before work startsvalidate_agent_actionApprove or block planned actionscheck_payment_requestEvaluate paid MCP or ASP callsscore_deliverableCheck delivery against criteriagenerate_dispute_packetPackage receipts and evidence
Registration data
NameIntentLockNetworkX Layer mainnetPrice$0.01 per paid callPay to0xec78...847D