tensorrt-llm
Pass
Audited by Gen Agent Trust Hub on Sep 21, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill documentation correctly describes the usage of the TensorRT-LLM library for inference optimization on NVIDIA GPUs.
- [EXTERNAL_DOWNLOADS]: References official containers from NVIDIA's Global Container Registry (NGC) at
nvcr.ioand documentation/source code from NVIDIA's official GitHub repositories. These are well-known and trusted technology services. - [INDIRECT_PROMPT_INJECTION]: The skill documents an interface for processing user-supplied prompts through LLM inference.
- Ingestion points: User prompts passed to the
llm.generate()function and the/v1/chat/completionsAPI endpoint. - Boundary markers: Absent in the basic usage examples.
- Capability inventory: The skill's scope is restricted to model inference; no dangerous capabilities such as arbitrary code execution or file system writes are connected to the prompt processing logic in the documentation.
- Sanitization: Not present in the simplified examples, which is standard for library documentation.
Audit Metadata