MANDATE EXECUTION LAYER

The permission layer between AI intent and money movement.

MandateShield is a shared execution boundary, not another checkout. Agents can discover it; applications can call it; payment systems can enforce its decision.

01User mandate

Amount, currency, merchant, purpose and expiry.

02AI agent

Finds an offer and prepares the final purchase facts.

03MandateShield

Returns ALLOW, REVIEW or BLOCK with exact reasons.

04Payment rail

Executes only after authority is confirmed.

HTTP API

One deterministic pre-flight request from any application or payment workflow.

POST /api/v1/preflight

MCP tool

A remote tool that AI clients can discover and call at the moment of purchase.

POST /api/mcp

A2A agent

A public agent card lets compatible agents delegate the authority decision.

/.well-known/agent-card.json

Open contract

Stable schemas, reason codes and receipts prevent every ecosystem from inventing a new boundary.

/standard

ONE CONTROL, EVERY AGENT

Integration without replacing the payment stack.

Credentials stay with the existing provider. MandateShield receives only normalized authority facts and produces a portable decision receipt immediately before execution.

Connect an application or agentDeveloper hub Implement the open boundary contractSpecification v1.0 Run a purchase through the boundaryFree live tool

START AT THE EXECUTION POINT

Add one decision before the payment call.

Connect MandateShield →