paperclip-create-agent

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

• Third-party content exposure detected (high risk: 0.80). The skill includes required templates (e.g., the QA "Browser Workflow" and SecurityEngineer/browser guidance in references/agents/*.md used as the mandated instruction source in step 4) that instruct agents to open target URLs and use a browser skill to read and verify web pages, which means the created agents will fetch and interpret arbitrary public third‑party pages (open URLs) as part of their runtime workflow.