Public execution cellRun the assurance chain.
Run the assurance chain.
Inspect what comes out.
This is the actual open-source STANDARD reference engine—not a video, static dashboard, or generated fixture. The campaign uses synthetic procurement data and cannot reach external tools.
01
Sealing subject, world, scenario, and seed manifests
02
Mediating tools, time, randomness, and external effects
03
Injecting stale data, permission loss, and prompt attacks
04
Evaluating oracles and estimating uncertainty
05
Signing evidence and evaluating the release gate
Procurement authority boundary
A real agent harness will process synthetic purchasing events while faults attempt to bypass policy.
STANDARD isolation · synthetic data · egress deniedWhat actually runs
Immutable manifests → seeded trials → mediated effects → causal traces → statistical measures → signed decision
Audit the source ↗Evidence semantics
What this run proves—and what it does not.
The exact sealed inputs produce harness-visible traces, measures, findings, and digests that can be replayed and independently verified.
Undeclared capabilities fail closed in the reference harness; statistical results include sample size, distribution, seed coverage, and uncertainty.
STANDARD reference isolation is not a microVM, HSM trust root, or claim of HIGH-RISK production conformance.
Passing simulation does not prove universal safety. A decision applies only to its subject digest, suites, operating envelope, conditions, and expiry.