The Agent Skills Directory

[SAFE]: The skill consists of documentation and configuration examples for established high-performance computing tools. No malicious behaviors or suspicious obfuscation were detected.
[NO_CODE]: The skill contains only markdown documentation and metadata files. It does not include any self-executing scripts, binaries, or automated tasks.
[EXTERNAL_DOWNLOADS]: The skill provides instructions for installing official software packages via standard package managers (pip) and pulling container images from verified registries. These operations target well-known repositories and are essential for the tool's primary purpose.
[PROMPT_INJECTION]: The skill documents interfaces for processing user-supplied prompts through inference engines, which represents an indirect prompt injection surface.
Ingestion points: User prompts are passed to the llm.generate method in code snippets and through trtllm-serve API endpoints.
Boundary markers: The provided examples do not include explicit delimiters or instructions to ignore embedded commands within user input.
Capability inventory: The skill enables high-throughput text generation and network-exposed model serving.
Sanitization: No input validation or sanitization logic is present in the documentation; users are expected to implement their own security layers when deploying these services.

tensorrt-llm