Narrator

• Verdict: Pass
• Confidence: Medium
• Analyzed: Feb 16, 2026, 9:55 PM
• Findings: 0
• Version: 1.3.2

Overview

The skill's stated purpose (live macOS screen narration using Gemini Vision + ElevenLabs TTS) matches the files, required binaries, and environment variables; nothing in the bundle indicates intentional misdirection, but the wrapper expects a local upstream checkout at a hard-coded path and will send screen data to third-party APIs, so review before use.

This skill appears to be a thin wrapper that delegates to a local repo at /Users/buddy/narrator and uses Google Gemini (vision) and ElevenLabs (TTS). Before installing or running: 1) Verify you actually have the upstream repo at /Users/youruser/narrator (the bundle expects /Users/buddy/narrator — adjust paths or the wrapper if needed). 2) Understand that screen contents will be sent to external services (Gemini/ElevenLabs) — evaluate privacy and API key scope. 3) Keep your GEMINI/ELEVENLABS API keys secret and consider creating limited-permission keys if possible. 4) Inspect the upstream repo referenced by the wrapper (not included here) to confirm no unexpected network endpoints or data exfiltration. 5) Run initially in a controlled environment (non-production user account) to confirm behavior, and check the /tmp control/status files and running tmux session before leaving it active.