runpod-cost-audit
Pass
Audited by Gen Agent Trust Hub on Nov 6, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill enforces a strict read-only policy, explicitly directing the agent never to stop, delete, or reconfigure resources autonomously.
- [INDIRECT_PROMPT_INJECTION]: While the skill processes data from external tool outputs (Runpod billing and resource lists), it possesses no high-privilege capabilities such as shell execution or file system modification that could be exploited via malicious data injection.
- [COMMAND_EXECUTION]: The instructions specifically prohibit the use of unauthorized tools like shell environments or web-fetchers, ensuring the agent operates within a restricted toolset provided for the audit task.
Audit Metadata