Does the boundary hold? Four plain-language checks, using made-up data and local reference rules. Each check runs three ways: a protected case, an allowed-action control, and a deliberately weak control.
What the results mean
4/4Protected cases passed. The restricted action was stopped (or, for excess access, flagged), as expected.
4/4Allowed-action controls passed. Permitted actions still went through, so the checks are not simply blocking everything.
4/4Deliberately weak controls failed — as expected. These are built to be weak. Their failure shows each check can detect a real gap.
SCOPELocal synthetic demonstrations only. These are not tests of vendor products (Airlock, Linx, Cato, Wiz: NOT TESTED) or of production enforcement, including the deployed Ethictecture site.
Local demonstration — vendor products NOT TESTED.
These results do not certify Airlock, Linx, Cato, Wiz, or your deployed Ethictecture. They show what this small reference harness did, in your browser, with fabricated data. The browser controls are editable demonstrations, not independent enforcement. A PASS on a blocked action means the observed result matched the expectation.
What remains to test
Actual product testing needs an authorized disposable environment, a pinned product version and configuration, and independent backend records. For each scenario, compare the decision log with the actual file, network, tool or permission outcome.
No private MINE Root implementation, personal records, vendor credentials or external calls are included. No results are sent anywhere. Reloading restores the packaged record; download a new run to keep it.
Scenario concepts: Brooke Broom / MINE lab. Code, synthetic fixtures and report generation: ChatGPT assistance. No claim of originality or vendor equivalence.