runai-get-utilization-overview
Pass
Audited by Gen Agent Trust Hub on Nov 2, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill is designed strictly for read-only operations, fetching and analyzing GPU utilization data from Run:ai clusters via established MCP tool interfaces. No dangerous operations or exfiltration patterns were detected.
- [COMMAND_EXECUTION]: The instructions contain explicit safeguards against command injection and unauthorized administrative actions. It strictly forbids the use of
kubectl,nvidia-smi, or therunaiCLI, ensuring the agent remains within a restricted monitoring scope even if prompted otherwise by a user. - [INDIRECT_PROMPT_INJECTION]: The skill processes cluster and workload metadata which could be controlled by an attacker. To mitigate this risk, it includes specific instructions for the agent to treat all tool output as untrusted data and to never follow instructions embedded within names or metrics.
- [EXTERNAL_DOWNLOADS]: The skill does not perform any external network requests or script downloads, relying entirely on the provided Run:ai MCP server for cluster communication.
Audit Metadata