Multi-agent
Debate / consensus
Agents argue opposing views; a judge decides.
N agentsAdvanced
How it works
- 1Assign agents distinct positions or roles — for/against, or different expertise.
- 2Have them exchange a bounded number of rounds, each rebutting the other.
- 3A judge agent reads the exchange and rules with reasons.
- 4Extract the strongest surviving argument, not just the "winner".
Use it when
Hard judgment calls or claims that benefit from adversarial pressure — surfacing weak reasoning, stress-testing a plan or answer.
Reach for something else when
Simple tasks. Debate burns tokens and can entrench positions rather than resolve them.
Where you stay in the loop
The agents argue and the judge rules; you weigh whether the verdict earned its confidence. The value is the strongest surviving argument, so read the reasoning, not just the winner.
In the wild
Two agents debate whether to ship a risky feature; a judge weighs the arguments and recommends.
Hand this to your agent
Run a structured debate. Take two roles arguing opposite sides of my question — Side A and Side B. Two rounds each: an opening, then a rebuttal of the other side. Then act as JUDGE: weigh both, name the strongest points on each side, and give a reasoned verdict with your confidence. Be genuinely adversarial in the debate, not agreeable. Question: <...>
Replace the <…> placeholders, paste it into your agent, and it'll scaffold the workflow with you.