runai-audit-assigned-quota
Pass
Audited by Gen Agent Trust Hub on Nov 2, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill instructions are designed with a strong security posture. They explicitly direct the agent to treat all tool outputs as untrusted data and to ignore any instructions potentially embedded within them (e.g., in workload pending reasons), which effectively mitigates the risk of indirect prompt injection.\n- [SAFE]: The skill is restricted to read-only operations using the Run:ai MCP tools. It contains explicit prohibitions against modifying cluster state, fabricating data, or falling back to shell commands (like
kubectl) if the MCP source is unavailable.\n- [SAFE]: The provided mock script (evals/files/mock_mcp.py) is used solely for evaluation purposes and contains only deterministic logic for replaying test data. It does not perform any dangerous operations such as network requests, file system modifications, or subprocess execution.
Audit Metadata