brainstorming

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

• Third-party content exposure detected (high risk: 0.80). The skill explicitly instructs delegating to an internet-researcher or to "use WebSearch/WebFetch aggressively" (see the "When to Use internet-researcher" and Research Protocol sections) to find docs, community patterns, and "how do people solve X" content on the open web, which means the agent will fetch and read untrusted third-party/user-generated web content.