The Agent Skills Directory

[SAFE]: The skill documents the use of a vendor-provided tool (doca_bench) located at a standard, expected system path (/opt/mellanox/doca/tools/doca_bench). It does not attempt to download, install, or execute arbitrary code from the internet.\n- [SAFE]: Explicit safety policies are included, specifically warning that the tool is not intended for production deployments and requiring a 'smoke-before-bulk' verification process to prevent resource exhaustion or misconfiguration during long benchmarking runs.\n- [SAFE]: All external references (URLs and repositories) point to official NVIDIA domains or well-known technology services, which is consistent with the skill's stated author (nvidia).\n- [SAFE]: The instructions emphasize using the tool's built-in help (--help) as the authoritative source for flags and parameters, which mitigates the risk of the agent using hallucinated or outdated commands.\n- [SAFE]: No evidence of prompt injection, data exfiltration, obfuscation, or persistence mechanisms was found in the provided instructions or metadata.

doca-bench