search

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

• Third-party content exposure detected (high risk: 0.70). SKILL.md describes a runtime workflow where the agent calls the Browserbase Search API and receives untrusted remote search result fields (e.g., title, url, author) that are free-text from third-party web content, which can be passed into the LLM context as part of “structured results.”