presentations

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

• Third-party content exposure detected (medium risk: 0.30). Outsider-authored free text is ingested at runtime via the required “Brand grounding” gate, which has the agent locate a brand study from root CLAUDE.md and, if absent/incomplete, STOP and “interview the user” and persist the user’s verbatim voice samples under 02-DOCS/raw/brand/ before proceeding.