k8s-envoy-gateway-troubleshooting
Pass
Audited by Gen Agent Trust Hub on Nov 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from the Kubernetes cluster, including container logs and resource status conditions, which could potentially contain malicious instructions aimed at influencing the agent's actions.\n
- Ingestion points: Commands such as
kubectl logsandkubectl get -o jsonpathinSKILL.mdand referenced files ingest external data into the agent's context.\n - Boundary markers: The skill relies on high-level instructions to mitigate risk, such as 'Read first, ask before changing' and categorizing all mutations under 'Ask first' or 'Proposed fix'. There are no technical delimiters surrounding the ingested content.\n
- Capability inventory: The skill possesses capabilities to modify the cluster state via
kubectl patch,kubectl apply, andkubectl rollout restart.\n - Sanitization: No technical sanitization, filtering, or validation is applied to the log or status data before it is interpreted by the agent.
Audit Metadata