brand-extract

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

• Third-party content exposure detected (medium risk: 0.65). The skill’s runtime path ingests outsider-authored free text from the target website’s DOM/CSS via agent-browser snapshot and then places it into brand.json (e.g., brand.voice copy and brand.tagline/description), which the renderer injects into the LLM context as part of the extraction/synthesis process.