Agentic AI is the tech industry’s latest hot topic, and companies like OpenAI and Anthropic are rushing to give businesses the tools to create these automated helpers. To keep up, OpenAI has now updated its agents software development toolkit (SDK). This update introduces several new features designed to help businesses create their own AI agents that run on OpenAI’s models.
The SDK’s new capabilities include a sandboxing feature. This allows AI agents to operate in secure, controlled computer environments. This is important because running agents without supervision can be risky due to their sometimes unpredictable behavior.
With this sandbox integration, agents can work in isolation within a specific workspace. They can access files and code only for particular operations, while otherwise protecting the system’s overall integrity. This creates a safer way to deploy AI.
Related to this, the new version of the SDK also gives developers an in-distribution harness for frontier models. This will allow agents to work with files and approved tools within a workspace, the company explained. (In agent development, a “harness” refers to the parts of an agent other than the main model it runs on. An in-distribution harness often lets companies deploy and test agents running on frontier models, which are the most advanced, general-purpose AI models available.)
Karan Sharma from OpenAI’s product team told TechCrunch, “This launch, at its core, is about taking our existing Agents SDK and making it so it’s compatible with all of these sandbox providers.” Sharma hopes that this, combined with the new harness capabilities, will enable users “to go build these long-horizon agents using our harness and with whatever infrastructure they have.” “Long-horizon” tasks are generally considered more complex, multi-step operations.
OpenAI plans to keep expanding the Agents SDK over time. Initially, the new harness and sandbox features are launching in Python, with TypeScript support planned for a later release. The company also said it is working to bring more agent capabilities, such as code mode and subagents, to both Python and TypeScript. The new Agents SDK capabilities are available to all customers through the API and will use standard pricing.
