absolute-brainstorm

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

• Third-party content exposure detected (medium risk: 0.30). Required workflow ingests outsider-authored free text via the user/attacker conversation content during the “Relentless interview” phases (e.g., Phase 4 “Use the AskUserQuestion tool for every question”), while also performing a local “Deep context scan” of the codebase (trusted within the tenant) for docs/README/commits.