exploration

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

• Third-party content exposure detected (high risk: 1.00). The SKILL.md explicitly instructs agents to autonomously search public sources (e.g., "searches arXiv for recent papers" and "explore competitor pricing pages") and ingest those third-party webpages as evidence that the agent reads and uses to make decisions, exposing it to untrusted public web content that could carry indirect prompt injections.