incident-response-commander
Pass
Audited by Gen Agent Trust Hub on Sep 10, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and act upon untrusted external data sources as part of its core incident management function.
- Ingestion points: The 'Workflow Process' section explicitly defines 'alert fires or user report received' as the primary ingestion points for the agent's tasks.
- Boundary markers: The instructions do not define explicit boundary markers or delimiters (such as XML tags or unique string markers) to separate these external inputs from the agent's instructional context.
- Capability inventory: While no tools are explicitly granted in the frontmatter, the skill body provides multiple templates containing significant infrastructure management commands (
kubectl rollout history,kubectl rollout undo,kubectl scale,kubectl restart). If the agent is equipped with a shell or terminal tool, these could be executed in response to potentially malicious alert data. - Sanitization: There are no instructions or patterns provided to validate, sanitize, or escape the content received from external alerts or user reports before they are processed by the agent.
Audit Metadata