Handoffs ↗

Editorial Notes

Handoffs are OpenAI’s primary mechanism for multi-agent orchestration — they let one agent explicitly transfer control to another, passing along context and intent classification. Compare this with LangGraph’s conditional edges (graph-based routing) and CrewAI’s task delegation (role-based assignment) to see three fundamentally different approaches to the same problem. A key gotcha is that handoffs are one-directional by default; if you need an agent to return results to the caller, you must explicitly configure return handoffs. Read the Agents doc first to understand how agent definitions work before layering handoff behavior on top.