The Agent Skills Directory

[SAFE]: The skill serves as a technical guide for NVIDIA TensorRT-LLM, providing legitimate code snippets for high-performance LLM inference and serving.
[EXTERNAL_DOWNLOADS]: The skill references official installation methods including pip install tensorrt_llm and docker pull nvidia/tensorrt_llm. These downloads target well-known repositories and official vendor images.
[COMMAND_EXECUTION]: Provides standard command-line instructions for environment setup, model serving, and benchmarking using tools like pip, docker, and the trtllm-serve CLI.
[DATA_EXFILTRATION]: Follows security best practices by instructing users to use environment variables ($API_KEY) for authentication rather than hardcoding sensitive credentials.
[INDIRECT_PROMPT_INJECTION]: While the skill demonstrates how to process user prompts through an LLM, it is a documentation-only skill and does not implement automated agents that process untrusted data without user oversight.

tensorrt-llm