Build Hour: Agents SDK

Why It Matters

The enhancements lower the barrier to deploying reliable, secure AI agents in production, accelerating enterprise adoption of autonomous workflows.

Key Takeaways

• Agents SDK now separates harness from compute for better scalability.
• First‑class sandbox support across Modal, Cloudflare, Docker, and local environments.
• New skills API lets versioned, reusable skill bundles be uploaded centrally.
• Hosted shell tool enables lightweight container execution via Responses API.
• Built‑in memory, web/file search, and tool orchestration simplify agent development.

Summary

OpenAI’s Build Hour session introduced the latest upgrades to the Agents SDK, a framework designed to let developers build long‑running, production‑grade AI agents without hand‑crafting orchestration logic. The presenters highlighted how the SDK now decouples the model‑native harness from the underlying compute, enabling agents to run in transient sandboxes while the harness manages state, snapshots, and rehydration. Key enhancements include first‑class sandbox integrations with platforms such as Modal, Cloudflare, Docker, and local environments, a new Skills API for uploading versioned skill bundles, and a hosted shell tool that spins up lightweight containers via the Responses API. Additional out‑of‑the‑box capabilities—agent memory, web and file search, and tool orchestration—reduce the engineering overhead of building complex workflows. Steve cited real‑world use cases: Codeex‑powered security agents scanning repositories for vulnerabilities, an internal data agent answering analytics queries, and a tax‑prep skill bundle that encapsulates IRS rules and scripts. The demo showed an agentic task tracker that leverages sandboxed execution, memory persistence, and skill reuse to complete multi‑step tasks autonomously. By abstracting sandbox management, secret handling, and state persistence, the updated SDK promises faster time‑to‑market for AI‑driven products, broader model‑agnostic flexibility, and stronger security guarantees for enterprises deploying agents at scale.

Original Description

Build with the next evolution of the Agents SDK. In this Build Hour, you’ll learn how to use the updated Agents SDK to build long-running agents with a model-native harness. Give agents the tools, memory, and execution environment they need to work across files and systems — while staying flexible enough to fit your own stack.

In this session, Steve Coffey (Member of Technical Staff) and Nish Singaraju (Product, Agents) will cover how to:

• Build agents that can inspect files, run commands, edit code, and operate across multi-step tasks

• Use the updated Agents SDK harness for more reliable agent loops across documents, tools, and systems

• Work with core agent primitives like MCP, skills, AGENTS.md, shell, and apply patch

• Run agents safely in controlled sandbox environments with the files, dependencies, and tools they need

👉 Agents SDK Docs: https://developers.openai.com/api/docs/guides/agents

👉 Typescript SDK: http://github.com/openai/openai-agents-js

👉 Python SDK: http://github.com/openai/openai-agents-python

👉 Agents SDK Quickstart: https://openai.github.io/openai-agents-python/quickstart/

👉 Follow along with the code repo: http://github.com/openai/build-hours

👉 Sign up for upcoming live Build Hours: https://webinar.openai.com/buildhours

00:00 Introduction

01:48 Agents SDK updates

11:21 Recent releases in the API

14:13 Demo: Building a task tracker with the Agents SDK

37:50 Q&A