research

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

• Third-party content exposure detected (high risk: 1.00). The skill's Research cycles explicitly instruct the agent to fetch and interpret public web content using Brave Search and Tavily Search (see "Initial landscape" and "Deep investigation" steps and the Tools listing), so the agent will read untrusted third-party material that can influence its analysis and actions.