@tank/local-llm-fit

Pass

Audited by Gen Agent Trust Hub on May 31, 2026

Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill utilizes scripts/detect_hardware.sh to execute system diagnostic commands such as sysctl, nvidia-smi, rocm-smi, and lspci. These commands are used to identify the host's CPU, GPU, VRAM, and memory specifications to perform fit calculations.
  • [EXTERNAL_DOWNLOADS]: The references/search-protocol.md instructions direct the agent to query well-known and trusted platforms, including Hugging Face, Ollama, and official model vendor websites (e.g., Meta, Mistral, DeepSeek), to retrieve technical model metadata.
  • [PRIVILEGE_ESCALATION]: The scripts/detect_hardware.sh script includes a command sudo -n dmidecode to attempt to retrieve RAM speed data. This is a non-interactive request for elevated privileges that is restricted to reading hardware metadata and is consistent with the skill's primary purpose of hardware diagnostics.
Audit Metadata
Risk Level
SAFE
Analyzed
May 31, 2026, 11:04 PM
Security Audit — agent-trust-hub — @tank/local-llm-fit