multi-llm

Pass

Audited by Gen Agent Trust Hub on Jun 22, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The documentation references the official installation URL for Ollama (https://ollama.com/install.sh). Ollama is a recognized service for local model execution, and its inclusion is a routine part of the setup for this skill.
  • [REMOTE_CODE_EXECUTION]: The SKILL.md file contains a command (curl | sh) for installing the Ollama platform. This pattern is noted as safe because the destination is a well-known technology service provider.
  • [COMMAND_EXECUTION]: The skill includes two shell scripts, select-model.sh and fallback-demo.sh, which execute locally to classify tasks and manage model selection. These scripts use standard unix utilities like curl, grep, and sed to interact with the local Ollama API at localhost:11434 and perform as described in the documentation.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 22, 2026, 05:32 AM
Security Audit — agent-trust-hub — multi-llm