Governance
Policies, guardrails and blocked actions, read from the services that enforce them.
Compliance
The AI registry and compliance packs (EU AI Act, NIST AI RMF, ISO/IEC 42001; SOC 2 and ISO/IEC 27001 evidence), filled from what this page shows, with exportable evidence bundles.
Your organization's AI registry
Organization admins register their own agents, models and the external AI systems they use, publish a notice for the people those systems affect, and handle the reviews of the decisions those systems take.
Blocked actions
Everything the platform refused to do, and why: policy denials, rejected tool-call approvals and PII guardrail refusals, read from the audit log. Each row's # is its audit log sequence number.
Loading blocked actions...
Pending approvals
Tool calls waiting for a person to approve or reject them, across every run.
Loading...
Rules in force
Read from the running services, not from the chart: the Model Gateway reports the guardrail config it loaded, and OPA reports the data it is deciding with.
Loading...
Known governance gaps
Controls this platform does not enforce yet. Each stays listed until its ticket is closed; NetworkPolicy until this install declares enforcement.
- VC-123
Harbor vulnerability scanning does not run, and cannot run air-gapped
No image in Harbor has ever been scanned for known vulnerabilities, and at an air-gapped site the scanner cannot fetch its vulnerability database.