verify-ai-server-with-device-env

Pass

Audited by Gen Agent Trust Hub on Jun 16, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses nvidia-smi, docker info, and ssh to inspect host driver versions and Docker runtime configuration.
  • [EXTERNAL_DOWNLOADS]: Troubleshooting guides reference the download of official GPG keys and setup scripts from NVIDIA's GitHub infrastructure for configuring the container toolkit.
  • [COMMAND_EXECUTION]: Provides instructions for the user to perform necessary system-level configuration using sudo for tasks like registering the container runtime and restarting the Docker daemon.
  • [PROMPT_INJECTION]: The skill analyzes external data in the form of service logs pasted by the user to identify model loading status and API health.
  • Ingestion points: Log output from docker compose provided by the user in Stage 4 (SKILL.md).
  • Boundary markers: None; the agent identifies status based on specific log markers.
  • Capability inventory: Ability to execute shell commands via the Bash tool and SSH connections.
  • Sanitization: No explicit sanitization of log content, though logic is limited to status verification.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 16, 2026, 08:26 AM
Security Audit — agent-trust-hub — verify-ai-server-with-device-env