Struct DebateAgent 

pub struct DebateAgent<P: LlmProvider + 'static> { /* private fields */ }

Orchestrates a multi-round debate between N debater agents, then asks a judge agent to synthesize the final answer from the complete transcript.

Each round, all debaters run in parallel and receive the full debate history as input. After all rounds (or early stop), the judge produces the final output.

Note on transcript growth: The full transcript is passed to every debater each round, so input tokens grow quadratically with round count and debater count. Keep max_rounds small (2–4) or use should_stop for early termination.

Implementations

impl<P: LlmProvider + 'static> DebateAgent<P>

pub fn builder() -> DebateAgentBuilder<P>

Create a new DebateAgentBuilder for constructing a debate agent.

pub async fn execute(&self, task: &str) -> Result<AgentOutput, Error>

Execute the debate.

Each round, all debaters receive the full transcript and run in parallel. After max_rounds (or early stop), the judge synthesizes the final answer.

Trait Implementations

impl<P: LlmProvider + 'static> Debug for DebateAgent<P>

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl<P: LlmProvider + 'static> From<DebateAgent<P>> for WorkflowRouter<P>

fn from(agent: DebateAgent<P>) -> Self

Converts to this type from the input type.