Cursor Launches Public Beta SDK So Developers Can Build With Its AI Agents

Cursor has opened up its coding agent technology to the public with the launch of the Cursor SDK, now available in public beta for all users.

The SDK lets developers take the same agents that power Cursor's desktop app and run them inside their own tools, pipelines, and products. Previously, these agents only worked inside the Cursor editor. Now they can run anywhere a developer needs them.

Using the SDK, a team can set up an agent to pick up a bug report, write a fix, and open a pull request, all without a developer stepping in. Agents can also run automatically when tests fail in a CI pipeline, summarize code changes, or help non-technical teams query data through a simple interface.

The package, available as @cursor/sdk on npm, is built for TypeScript. Developers can run agents locally on their own machine or on Cursor's cloud, where each agent gets its own virtual machine with the repository already cloned. Cloud agents keep running even if the connection drops, which makes them suitable for longer or more complex tasks.

Companies including Rippling, Notion, Faire, and C3 AI are already using the SDK in their workflows.

Cursor has also released a public cookbook on GitHub with ready-to-use starter projects, including a kanban board that manages cloud agents and a simple command-line tool for spinning up agents from a terminal.

Full documentation is available at Cursor Docs, and the blog post from the Cursor team walks through the full feature set in detail.