page-langs

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

• Third-party content exposure detected (high risk: 0.85). Yes—Step 2/3 uses playwright-cli open "$URL" and playwright-cli run-code to extract the target webpage’s DOM text and language signals (outsider-authored free text from an arbitrary URL) into pageData.text/pageData.*, which scripts/detect.mjs then processes and writes to langs.json (and thus becomes LLM-readable output context if the agent reads it).