The Agent Skills Directory

[SAFE]: The skill provides legitimate technical documentation and configuration guides for the NVIDIA TensorRT-LLM library and contains no malicious behavior, prompt injections, or obfuscation.
[EXTERNAL_DOWNLOADS]: The skill provides instructions to download the official NVIDIA TensorRT-LLM Docker image and the library via the Python package registry (PyPI).
[EXTERNAL_DOWNLOADS]: Model weights are downloaded from HuggingFace (e.g., Meta-Llama models) during the intended model initialization process, which is standard behavior for LLM tooling.
[COMMAND_EXECUTION]: Includes standard command-line examples for deploying inference servers, running benchmarks, and monitoring hardware status with nvidia-smi.

tensorrt-llm