ops-runpod

Pass

Audited by Gen Agent Trust Hub on Aug 26, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes local shell commands via the subprocess module to generate SSH key pairs (ssh-keygen) which are required for secure communication with remote RunPod instances.\n- [COMMAND_EXECUTION]: Uses the paramiko library for extensive remote shell command execution on user-provisioned instances to perform environment setup, training job management, and performance monitoring.\n- [EXTERNAL_DOWNLOADS]: Interacts with official RunPod GraphQL and REST endpoints to provision and terminate GPU infrastructure.\n- [EXTERNAL_DOWNLOADS]: The skill initiates downloads of model weights from HuggingFace and installs Python packages from PyPI (e.g., vllm, unsloth, torch) on remote instances to enable training and inference workloads.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 26, 2026, 06:00 PM
Security Audit — agent-trust-hub — ops-runpod