agnes-ai-generation

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

• Potentially malicious external URL detected (high risk: 0.80). The scripts make runtime calls to the API host https://apihub.agnes-ai.com (BASE_URL) and use returned chat-completion content (e.g., the translate_prompt_to_english output) as prompts for subsequent image/video generation, so externally-fetched content directly controls agent prompts.