page-cro

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

• Third-party content exposure detected (high risk: 0.75). The required workflow can fetch arbitrary public web content at runtime via scripts/conversion_audit.py (--url → urllib.request.urlopen(args.url) → html = resp.read().decode(...)), and that fetched page text is then parsed into full_text and used to generate the audit output that the agent would incorporate into its LLM context (indirect prompt-injection risk from outsider-authored page body text).