auto-loom

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

• Potentially malicious external URL detected (high risk: 0.90). scripts/compose.mjs downloads and saves a remote JS file at runtime from https://cdn.jsdelivr.net/npm/gsap@3.14.2/dist/gsap.min.js and then includes/executes it in the generated composition (and also invokes npx hyperframes to run a remotely-resolved package), so the skill fetches and executes remote code during runtime.

HIGH W008: Secret detected in skill content (API keys, tokens, passwords).

• Secret detected (high risk: 1.00). I flagged a single literal, high-entropy API key in .env.local. It is not a placeholder (it begins with "sk_" and is a long random-looking string), so it meets the secret definition and appears to be an active ElevenLabs API key. I ignored the placeholder in .env.example (ELEVENLABS_API_KEY=sk_your_key_here) as a documentation placeholder, and I did not treat listed voice IDs (ELEVENLABS_VOICE_ID) as secrets.