multi-debate

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

• Third-party content exposure detected (high risk: 1.00). The skill explicitly requires listing and reading peer-written files from a user-specified shared folder (see SKILL.md: "List the folder. Read every turn-{N-1}-*.md file except your own"), which are untrusted user/agent-generated third-party contents that the agent must interpret and use to revise its answers.