The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses shell commands for benchmark compilation, running evaluations, and containerizing benchmarks with Docker. These are standard operations for the tool's intended purpose.
[DYNAMIC_EXECUTION]: The skill generates Python benchmark files and uses pip to install requirements specified in the benchmark configuration. It also performs smoke tests on generated code using python3 -c to ensure correctness before deployment.
[EXTERNAL_DOWNLOADS]: The skill fetches datasets from HuggingFace (hf://) and references official NVIDIA API endpoints. These interactions target well-known and vendor-owned services.
[DATA_EXPOSURE]: The skill accesses local data files (JSONL, CSV) provided by the user to identify schemas and generate evaluation prompts. This access is required for its core functionality of benchmark creation.

byob