Skills
skills/gaojizhou/skills/local-llm-privacy/Gen Agent Trust Hub

local-llm-privacy

Pass

Audited by Gen Agent Trust Hub on Mar 14, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONREMOTE_CODE_EXECUTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill references installation procedures from ollama.com, which is a well-known service for local LLM inference. This includes downloading the official installer and pulling models through the Ollama CLI.
  • [COMMAND_EXECUTION]: The agent uses standard system commands such as ollama list and ollama serve to check the status of and start the local model service.
  • [COMMAND_EXECUTION]: Interacts with the local Ollama REST API via curl on localhost:11434 to perform text generation and chat tasks.
  • [COMMAND_EXECUTION]: Utilizes the base64 command-line tool to encode local image files for processing by vision-capable local models.
  • [REMOTE_CODE_EXECUTION]: The skill provides a standard installation command (curl | sh) for Linux users that fetches a script from the official well-known service domain ollama.com.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 14, 2026, 12:26 PM