qa

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

• Third-party content exposure detected (high risk: 0.85). The workflow’s Phase 2 “Fetch the origin page via HTTP” pulls arbitrary outsider-authored web page HTML from each _source_url at runtime, and then “Parse both … into a content model” feeds that fetched prose into the agent’s LLM context for diffing/grading.