High-level orchestration strategy for multi-agent runs.
"sequential" — agents are called one after another; output of each feeds the next.
"parallel" — all agents are invoked concurrently; results are merged.
"debate" — agents iteratively argue and refine a shared answer.
"review-loop" — one agent produces output, another reviews and requests revisions.
"hierarchical" — a coordinator agent dispatches sub-tasks to specialist agents.
"graph" — explicit dependency DAG; agents run when all dependsOn predecessors complete.
"panel" — every seat answers the same prompt on its own model; healthy seats are counted against a quorum and a chair synthesizes (see AgencyOptions.modelPool, chair, panel).
High-level orchestration strategy for multi-agent runs.
"sequential"— agents are called one after another; output of each feeds the next."parallel"— all agents are invoked concurrently; results are merged."debate"— agents iteratively argue and refine a shared answer."review-loop"— one agent produces output, another reviews and requests revisions."hierarchical"— a coordinator agent dispatches sub-tasks to specialist agents."graph"— explicit dependency DAG; agents run when alldependsOnpredecessors complete."panel"— every seat answers the same prompt on its own model; healthy seats are counted against a quorum and a chair synthesizes (see AgencyOptions.modelPool, chair, panel).