The OpenAI Agents SDK is a lightweight framework built around agents, handoffs between agents, guardrails that check inputs and outputs, and built-in tracing. Its human-in-the-loop support can pause a run for approval before sensitive tool calls, and it can attach MCP servers as tool sources.
Read in this order
- OfficialOpenAI Agents SDK documentationStart here: the core concepts of agents, handoffs and tools in one place.
- OfficialGuardrails (OpenAI Agents SDK)How input and output checks work, and what they can and cannot catch.
- OfficialHuman-in-the-loop (OpenAI Agents SDK)How to pause a run for approval before a sensitive tool call.