chatgpt-images-fallback

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

• Potentially malicious external URL detected (high risk: 0.80). The skill's runtime Playwright automation explicitly opens and interacts with https://chatgpt.com/images (see scripts/run_chatgpt_fallback.mjs and generate_with_fallback.py), using that external site to execute remote model code to generate images and relying on it as a required fallback—so the URL is a runtime dependency that influences outputs.