Close coverage gaps

Coverage shows capabilities by source or scope, with a status, evidence detail, suggested next step and evaluation time. Use it before claiming an organization or repository was assessed.

Synthetic demo data. Review your own source-level status and evaluation time before drawing a conclusion.

Read the summary

The header reports how many supported capabilities are fully evaluated. A partial or outdated capability is not counted as fully evaluated. Unsupported capabilities are displayed separately so they cannot make the denominator look artificially complete.

Source configuration evaluations become outdated after the application’s freshness threshold, currently seven days. Runtime telemetry uses a shorter freshness window, currently one day without events. These labels are operational signals; the right review interval for your organization may be shorter.

Work from the gap

You see	Next step
No source connected	Connect GitHub in Integrations.
Connected, not scanned	Run discovery or the relevant passive assessment.
Permission denied	Correct installation or repository access, then rerun.
Partial	Read the gap detail, fix that prerequisite, and repeat the evaluation.
Outdated	Verify the source is healthy and collect current evidence.
No runtime telemetry	Create an integration and instrument a sender.
Remote server not probed	Review scope and authorization in the assessment wizard.
Unsupported	Use another review method; no ModelFence test has established a result.

Keep scope with the decision

When reporting findings or a remediation, include the organization, source run and evaluation time. An asset with no findings but missing runtime telemetry has a different evidence boundary from one with both repository analysis and current observed activity.

Coverage is a statement about available evidence, not a compliance certification, security rating or guarantee that all AI services have been discovered.