jetson-llm-benchmark
Pass
Audited by Gen Agent Trust Hub on Jun 22, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The
bench_ollama.shscript references the official Ollama installation script (https://ollama.com/install.sh) within an informative error message designed to assist the user if the tool is not found. Ollama is a well-known service provider in the AI ecosystem. - [COMMAND_EXECUTION]: The skill uses shell scripts to run performance measurement tools via Docker containers (
vllm/vllm-openai,ghcr.io/nvidia-ai-iot/*) and local CLI utilities. These operations are required for the skill's primary purpose of benchmarking hardware. - [SAFE]: The skill identifies and interacts with local LLM server endpoints (e.g., localhost) and collects hardware telemetry (e.g., tegrastats) to generate performance reports. No evidence of credential theft, data exfiltration, or malicious prompt injection was detected.
Audit Metadata