generate-minecraft-3D-build-plan

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.70). This skill sends user-provided prompts and arbitrary HTTP/HTTPS image URLs to the external OneKey agent endpoint at https://agent.deepnlp.org/agent_router and parses/uses the agent's JSON response, so it clearly ingests untrusted third‑party content (user-supplied URLs and remote agent output) that can influence generated build plans.