skills/nvidia/skills/doca-caps/Gen Agent Trust Hub

doca-caps

Pass

Audited by Gen Agent Trust Hub on Jul 18, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill's primary purpose is system observability using the trusted doca_caps binary located at the absolute path /opt/mellanox/doca/tools/doca_caps. This binary is part of the official NVIDIA DOCA SDK installation.
  • [SAFE]: The instructions explicitly enforce a read-only and side-effect-free policy, instructing agents to avoid state-changing commands or workarounds when diagnosing capability discovery.
  • [SAFE]: All external references, such as docs.nvidia.com and GitHub repositories under the NVIDIA-DOCA organization, are verified vendor resources.
  • [SAFE]: The skill includes a digital signature (skill.oms.sig) and evaluation metadata (BENCHMARK.md, evals/evals.json) which are standard artifacts for verified skills and do not contain hidden malicious logic.
  • [SAFE]: Data exposure is limited to hardware attributes (PCIe addresses, MAC addresses) required for the skill's intended discovery function. The skill correctly routes sensitive tasks like installation to the appropriate setup skills.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 18, 2026, 05:55 AM
Security Audit — agent-trust-hub — doca-caps