Agent SDK

Deploy autonomous, multi-step AI agents fast: the Agent SDK provides a lightweight, Python-first framework for LLM orchestration, tool integration, and enterprise-grade tracing.

The Agent SDK is your production-ready framework for building complex, autonomous AI agents. It simplifies LLM orchestration (supporting OpenAI, Anthropic, and others) by managing the full execution loop, tool invocation, and multi-agent handoffs. Developers use Python functions to create tools (e.g., `WebSearchTool`, `CalendarTool`), which the agent autonomously calls to complete tasks. Key features include built-in tracing for debugging and optimization, plus configurable Guardrails for input/output validation. This is the efficient path to deploying reliable agents in production.

https://github.com/openai/openai-agents-python