browser-act-skill-forge

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

• Third-party content exposure detected (low risk: 0.10). Runtime workflow reads outsider-authored free text from the user-provided “business intent” (Phase 1 parses user input) and then uses it to drive page exploration/search and extraction, but the actual LLM ingests only that user text—not first-party site/applicant feeds like email/inbox/tickets or arbitrary externally submitted content without selection.