agent-council

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 the CLI question argument (e.g., agent-council "<text>"), which is passed at runtime into run_debate and then injected into every LLM phase prompt (premortem, position, cross-examination, and synthesis) without any required selection of a specific item.