The trust control plane
for machine execution.
Constrain what agents can do. Attest what they did. Verify it anywhere.
The missing layer in the agent stack.
Agents can discover tools, interoperate across runtimes, and initiate payments. But no neutral layer answers the three questions every enterprise, protocol, and counter-party needs answered.
What was allowed?
Who authorized this agent to act, and under what constraints?
What happened?
What did the agent actually execute, and is there cryptographic proof?
What can be verified?
Can a third party independently confirm the outcome without trusting the executor?
Where xMandate sits.
A neutral trust layer between agent runtimes and payment rails — not replacing either, verifying both.
Who initiates actions and under what authority
Where settlement and delivery happen
From policy to proof in three steps.
Every agent action follows the same trust lifecycle — declare what is allowed, produce cryptographic proof of what happened, verify independently.
Built for the surfaces that matter.
xMandate sits above protocols and runtimes as a neutral trust layer.
Agent Commerce
Attestation before settlement for machine-native payments — x402, MPP, and any payment rail where proof of execution matters.
Tool-Using Agents
Policy enforcement and receipts for every tool invocation.
Multi-Agent Systems
Verification primitives for cross-agent trust.
Internal Workflows
Audit trails and governance for internal agent deployments.
xmandate-sar-formatter
An independent TypeScript SAR implementation, available on GitHub and npm. Validated against the published fixtures and shared in the SAR proposal discussion.
- Published on npm
- Portable cryptographic receipts
- Deterministic receipt identity
- Validated against published SAR fixtures
- Offline verification
- Node and edge runtime compatible