Our AI built a working exploit and we independently verified it fires against the target in an isolated lab.

The run completed cleanly but the CVE was not reproducibly exploitable in-lab — an honest negative, not an error.

The assessment errored out or exhausted its attempts before reaching a verdict (lab build / infrastructure / harness failure), so exploitability is undetermined.

The vulnerability can't be assessed by our Linux-container pipeline (e.g. Windows/macOS-only, hardware/appliance, VMware, or purely client-side) and was not attempted.

We couldn't derive reliable version / target information to build a faithful lab, so the CVE was skipped rather than run.