runpod-pod-doctor
Pass
Audited by Gen Agent Trust Hub on Nov 6, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill follows security best practices by implementing a read-only diagnosis phase and requiring explicit user authorization before proceeding to recovery actions. No malicious patterns or exfiltration vectors were detected.
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted pod logs but mitigates risk through design. 1. Ingestion points: Pod logs via stream-pod-logs in SKILL.md. 2. Boundary markers: Clear separation of diagnosis and recovery phases; explicit instruction to ignore mutation commands in logs. 3. Capability inventory: Uses pod-action, delete-pod, and create-pod tools. 4. Sanitization: Requires human-in-the-loop authorization for all destructive or billable actions.
Audit Metadata