Runtime authority + signed evidence, in eight modules.

Every actor — human, agent, sub-processor, federated principal — gets a KYE™ URN. Lifecycle, classification, trust-domain. The evidence layer every other module binds to. Read more →

The runtime decision point. Every delegated action is evaluated against the bound authority chain at the moment it's proposed — before it crosses the commit boundary. Read more →

Every action cites a declared purpose. Every grant carries a bounded scope. Every decision is admissible against those bindings — or it doesn't reach commit. Read more →

For every runtime decision: a signed structural record of inputs, rules consulted, signals fused, verdict, and obligations. The contemporaneous answer to "what did the AI decide and why".

The canonical artefact a regulator, dispute panel or third-party auditor consumes. Replayable from the publisher's JWKS alone — no vendor dependency in the audit path. Read more →

Banking-grade replay: anyone with the publisher's JWKS can re-derive the decision offline. No vendor server in the verification path. The construction is proprietary.

What data an agent touched, why it touched it, whether the purpose was valid, what evidence proves it. The KYE Data Governance Pack™ ships this end-to-end.

Compiled authority bundles deployable to the edge. The Compiled Authority Bundle™ + Edge Arbiter™ + Safety Floor™ + Offline Evidence Log™ ship governance into network-partitioned, regulated, or sovereign deployments.