gemini-designer

CRITICAL E005: Suspicious download URL detected in skill instructions.

• Suspicious download URL detected (high risk: 0.80). Both URLs point to third‑party code and an unknown API: the GitHub repo is an unverified user’s package that would be installed and run locally, and the CLI is configured to talk to the nonstandard API endpoint https://zenmux.ai/api/v1 which could receive API keys, files, or exfiltrate data.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 1.00). The CLI calls the external provider at https://zenmux.ai/api/v1 (used at runtime in request_gemini) to fetch model completions that directly determine the agent's output, and the skill requires that endpoint to operate.

MEDIUM W013: Attempt to modify system services in skill instructions.

• Attempt to modify system services in skill instructions detected (high risk: 0.80). The skill explicitly instructs the agent to run a bundled installer script ("scripts/install_cli") silently and modify the CLI installation/path, which directs the agent to change the local system state (potentially requiring elevated privileges and altering PATH/config), so it should be flagged.