OpenAI updates its Agents SDK to help enterprises build safer, more capable agents

Agentic AI is the tech industry’s latest success story, and companies like OpenAI and Anthropic are rushing to give businesses the tools they need to create these automated little helpers. To this end, OpenAI has now updated the software development toolkit (SDK) for agentsintroducing a number of new features designed to help companies create their own agents running on the back end of the OpenAI models.

The SDK’s new capabilities include a sandbox capability, allowing the agents to work in controlled computing environments. This is important because agents can work completely unattended risky due to their sometimes unpredictable nature.

Sandbox integration allows agents to work in silos within a given workspace, only accessing files and code for certain operations, while otherwise protecting the overall integrity of the system.

Related to this, the new version of the SDK also provides developers with an in-distribution harness for boundary models that allows these agents to work with files and approved tools within a workspace, the company said. (In agent development, the “harness” is a term that refers to the other components of an agent besides the model it runs on. An in-distribution harness often allows companies to both deploy and test the agents running on boundary models, which are deemed to be the most advanced general purpose models available.)

“At its core, this launch is about taking our existing Agents SDK and making it compatible with all these sandbox providers,” Karan Sharma, who works on OpenAI’s product team, told TechCrunch.

The hope is that this, combined with the new harness capabilities, will allow users to “start building these long-horizon agents using our harness and whatever infrastructure they have,” he said.

Such long-horizon tasks are generally considered more complex, multi-step work.

OpenAI said it will continue to expand the Agents SDK over time, but initially the new harness and sandbox capabilities will launch in Python first, with TypeScript support planned for a later release. The company said it is also working on adding more agent capabilities, such as code mode and sub-agents, to both Python and TypeScript.

The new Agents SDK capabilities are offered to all customers via the API and use standard pricing.