← Back to all projects

OpenAI Agents SDK
A deliberately small agent framework from OpenAI
OfficialMIT
What is OpenAI Agents SDK?
Few primitives — agents, handoffs, guardrails, tracing — and little else. The small surface area is the appeal: there is not much to learn and not much to fight. It is built around OpenAI's own models first, so weigh that if multi-provider portability matters to you.
What can you do with OpenAI Agents SDK?
- Delegate between specialist agents — Handoffs and agents-as-tools let one
Agentpass work to another, andRunner.run_syncexecutes the whole chain and returnsfinal_output. - Give an agent a workspace —
SandboxAgenttakes aManifestof entries such asGitRepoand runs commands, inspects files and applies patches throughUnixLocalSandboxClient, orDockerSandboxClienton Windows. - Build voice agents two ways —
RealtimeAgentholds a WebSocket session withgpt-realtime-2.1, whileVoicePipelinechains speech-to-text, an agent workflow and text-to-speech from the optionalvoiceextra. - Point it at models other than OpenAI's — The README calls the SDK provider-agnostic across the OpenAI Responses and Chat Completions APIs plus 100+ other LLMs, though every quickstart example still expects
OPENAI_API_KEYin the environment. - Trace and resume conversations — Tracing is built in for viewing and debugging runs, and
Sessionscarries conversation history across separateRunnerinvocations without manual bookkeeping.
Star history
15 Aug to 30 Oct · +1.7k
28.7k30.4k
Frequently asked questions
Is OpenAI Agents SDK free for commercial use?
OpenAI Agents SDK is released under the MIT licence — OSI-approved open source, which permits commercial use.
How can OpenAI Agents SDK be deployed?
OpenAI Agents SDK is available as Self-hosted / Runs locally.