The Agent Skills Directory

[SAFE]: The skill is a documentation-only resource with no executable scripts or code. It provides structured guidance for using the pre-installed doca_bench utility.
[COMMAND_EXECUTION]: The skill instructs the agent to invoke the doca_bench binary and related system utilities (e.g., mlxconfig, numactl, dmesg) for performance tuning and verification. These are standard operations for the tool's intended use case on DOCA-enabled systems.
[EXTERNAL_DOWNLOADS]: The skill references official NVIDIA documentation and GitHub repositories. These are trusted vendor resources and do not involve untrusted third-party code.
[DATA_EXFILTRATION]: The skill describes network communication between the benchmark harness and a companion app for remote testing scenarios. It includes explicit safety warnings regarding the sensitivity of this channel and advises against use on production network segments.
[PROMPT_INJECTION]: No prompt injection, jailbreak attempts, or instructions to override safety filters were detected.
[OBFUSCATION]: No obfuscated strings, encoded commands, or hidden characters were found in the skill files.

doca-bench