pbl-moderator

To coordinate speaker rotation in a multi-agent PBL discussion, this skill enforces fair turn-taking and role-neutral facilitation by evaluating discussion state, history, and state cards to select the next speaker.

The best way to decide when to end a PBL discussion is by evaluating material coverage, question status, and participant readiness. This skill applies decision-driven ending policies to make disciplined, auditable choices.

Yes, you can manage multi-agent OpenCode sessions without generating dialogue. The moderator focuses strictly on enforcing turn-taking and making decisive choices, leaving content generation to participating agents.

A discussion moderator evaluates discussion state by analyzing material coverage, question status, and participant readiness. It uses these factors alongside discussion history and state cards to determine the next speaker or end the session.

A limitation when using state cards to manage discussion flow is that the moderator requires strict JSON-structured results and adherence to safety constraints, meaning it cannot operate outside its defined decision policy and hard rules.