deep-plan

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

• Third-party content exposure detected (high risk: 0.75). Outsider free text can enter the LLM context via Phase 1’s internal /research workflow (invoked as Skill(research, args="<topic>") in orchestrator mode, or executed inline in Teams mode), which commonly fetches/reads public web content and other non-user-authored sources at runtime.