brainstorm

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

• Third-party content exposure detected (high risk: 0.90). This SKILL.md explicitly tells the agent to "do relevant research on web" and the examples show "Searches the internet for similar props," meaning the agent will fetch and interpret open/public web content (untrusted user-generated sources) as part of its workflow, which can influence decisions and enable indirect prompt injection.