matter-intake-scoping-scott-margetts

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

• Third-party content exposure detected (high risk: 0.85). The skill’s runtime LLM context is populated by Mode 1/4 “client-provided inputs” (e.g., client emails, call notes, prior transaction files, data room index contents) that are outsider-authored free text, which the workflow instructs to extract and source/cite into the brief.