crw-extract

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

• Third-party content exposure detected (high risk: 0.85). Outsider free text can enter the LLM context when the required runtime workflow uses crw scrape/crw_crawl with formats:["json"] (or POST /v2/extract) to fetch and extract from a user-supplied URL/page, i.e., public web content scraped at runtime becomes readable prose for the server-side LLM.