Supervisor agent
An agent that owns a task and calls specialist agents for bounded subtasks, then combines their results.
A supervisor agent is an industry term for an agent that keeps control of a task and calls specialist agents for bounded subtasks, then combines their results. The OpenAI Agents SDK calls this pattern agents as tools and recommends it when one agent should own the final answer or enforce shared guardrails in one place; other frameworks describe a manager or coordinator in the same role. Because the supervisor is itself a model making delegation decisions, its choices need logging, and money-moving tools are best kept away from it.
Agent Minute explains this term on 7 March 2027.
Related terms
Multi-agent systemSeveral agents that divide a task between them and exchange work through a protocol.Agent handoffPassing control of a conversation from one agent to a specialist agent, which then takes over the turn.Agent frameworkA software library for building and running agents: the model loop, tools, delegation, approvals and tracing.Excessive agencyGiving an LLM application more functions, permissions or autonomy than needed, so bad outputs cause harmful actions.