Glossary
The words these docs use. Where a term names something built or proposed, its status is beside it.
Security terms
| Term | What it means | Where it stands |
|---|---|---|
| Artifact integrity | The model bytes are checked against the approved digest before a session loads them. | In the experiment |
| Configuration integrity | The configuration digest is part of the release and compared on every report. | In the experiment |
| Release authorization | A release is signed and approved before it can be loaded. | In the experiment |
| Destination-bound approval | An approval names the environment and resource it applies to. | In the experiment |
| Runtime identity | The runtime reports which release it loaded, under which runtime policy. | Worker self-report |
| Version mismatch | Approved and reported identities differ, shown as reload pending. | In the experiment |
| Deployment evidence | Every observation is kept, including ones that confirm nothing. | Recorded trace |
| Recovery verification | After an explicit reload the reported identity is compared again. | One recorded reload |
| Unauthorized changes | Wrong destinations and changed components as scripted fault tests. | Specified, not run |
| Evidence freshness | Heartbeats that show a report is current, not stale. | Not built |
| Model provenance | Where a model and its training inputs came from, beyond its digest. | Research |
Release and evidence terms
| Term | What it means |
|---|---|
| Release | One exact thing, named by digest: the model, its configuration and, proposed, its preprocessing. Release identity |
| Destination | Where a release may run. In the captured run, the environment non-actuating-lab and the resource onnx-cpu. |
| Approval | A binding of one exact release to one destination, under a runtime policy with an age limit. Approval binding |
| Runtime policy | The runtime a destination must have. In the experiment, ONNX Runtime 1.23.2 with exactly CPUExecutionProvider. Runtime integration |
| Observation | One recorded comparison of the approved identity with the identity the runtime reports, with its status. Evidence and observations |
| Matched | Every identity field is equal, and the reported runtime satisfies the policy. |
| Reload pending | The reported identity differs from the approved one. In baseline mode the runtime waits for an explicit reload. |
| Evidence unavailable | There is no loaded report. The observation carries a reason code and no loaded identity. |
| Reason code | Why an observation has the status unknown. The captured run has one, RUNTIME_UNAVAILABLE. |
| Self-report | What the worker says it loaded. The host and worker are trusted in the experiment, so it is not independent attestation. |
| Baseline mode | The runtime keeps serving the release it loaded until an explicit reload. |
| Trace | One JSON document holding a run's scope, recorded toolchain, runtime policy and observations. |
| Imported, not verified | The label on every value from a trace opened in the evidence app. Signatures are not checked. The evidence app |
Kinds of statement
Working in the experiment, measured, proposed, hypothesis and not claimed. How to read these docs
Last updated: October 10, 2026