debug

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

• Third-party content exposure detected (high risk: 0.70). The required workflow includes “External Research” (Track 3) to “search for known issues” and “find similar bugs in dependency issue trackers,” which at runtime typically involves fetching and summarizing outsider-authored web content/messages that can become free text in the agent’s LLM context.