deep-learner

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

• Third-party content exposure detected (high risk: 1.00). The skill's Step 1 explicitly says "If user provides a URL → use WebFetch to retrieve content," meaning the agent fetches and ingests arbitrary public/user-provided web pages (untrusted third‑party content) and reads/interprets them as part of its workflow.