baoyu-comic

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

• Potentially malicious external URL detected (high risk: 1.00). The skill calls the OpenAI chat completions API at runtime (defaulting to https://api.openai.com/v1/chat/completions or a custom OPENAI_API_BASE/OPENAI_BASE_URL), and the LLM response JSON (visual_prompt fields) is parsed and used directly as the prompts/instructions for image generation, so external content from that URL directly controls agent prompts.