tensorrt-llm
Pass
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: SAFE
Full Analysis
- SAFE (SAFE): No malicious patterns or security risks were detected across the analyzed files. The skill correctly instructs users to install dependencies from established registries and provides legitimate configuration examples.
- Indirect Prompt Injection (SAFE): The skill defines patterns where untrusted user data is ingested for inference, creating an indirect prompt injection surface. However, the risk is negligible as the skill lacks exploitable capabilities (such as shell access or file writing) that could be triggered by malicious prompt output.
- Ingestion points: User prompts ingested in
SKILL.md(Basic inference) andreferences/serving.md(FastAPI and curl examples). - Boundary markers: Absent; prompts are passed directly to the model.
- Capability inventory: Limited to model inference and serving; no evidence of subprocess spawning or file system modification based on model output.
- Sanitization: Absent; standard for raw inference wrappers.
Audit Metadata