NYC

local-llm-ops

Warn

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [COMMAND_EXECUTION] (MEDIUM): The skill instructs the user to execute multiple local shell scripts (setup_chatbot.sh, install_aliases.sh, scripts/run_benchmarks.sh, diagnose.sh) and Python scripts (chat.py). Since the contents of these scripts are not included in the skill definition provided, their behavior cannot be verified for malicious commands.
  • [EXTERNAL_DOWNLOADS] (LOW): The skill recommends installing the ollama package via Homebrew (brew install ollama). While Homebrew is a reputable source, this constitutes an external software dependency.
  • [DATA_EXFILTRATION] (SAFE): Network operations are limited to localhost:11434 for communicating with the local Ollama service, which does not pose an exfiltration risk.
  • [PERSISTENCE] (LOW): The use of install_aliases.sh suggests modification of shell profiles (e.g., .bashrc or .zshrc) to create aliases, which is a form of environment persistence.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Feb 17, 2026, 05:31 PM