production-troubleshooting
Pass
Audited by Gen Agent Trust Hub on Sep 23, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data from logs and traces, which could be used to deliver malicious instructions to the agent. \n
- Ingestion points: Application logs are retrieved using
k8s-tool logsandkubectl logs, and Sentry traces are analyzed in Step 2B and Step 3 ofSKILL.md. \n - Boundary markers: The skill does not define specific delimiters or warn the agent to ignore instructions embedded in logs or traces. \n
- Capability inventory: The agent has shell execution capabilities via
kubectlandk8s-tooland is instructed to modify Helm configuration files (SKILL.md, Step 5). \n - Sanitization: No sanitization or validation of log/trace content is performed before processing. \n- [COMMAND_EXECUTION]: The skill documents the use of potentially sensitive commands for environmental investigation and modification. \n
- Evidence:
references/common-issues.mdincludes commands for executing processes within pods (kubectl exec) and restarting system services (kubectl rollout restart).
Audit Metadata