The AI-powered Cursor code editor is now opening its core technology to all developers. Cursor has announced the release of the public Beta version. Cursor SDK — a TypeScript library that gives engineers programmatic access to the same runtime, harness, and models that power Cursor’s desktop app, CLI, and web interface.
The AI-coding tool market is changing dramatically. It’s no longer just an interactive assistant that sits alongside the developer. Instead, it has evolved into a deployable infrastructure which organizations can integrate with their current systems.
The Interactive Tool and Programmable Infrastructure
If you’ve used Cursor before, you know it as an IDE where you interact with an agent in real time — asking it to write functions, fix bugs, or explain code. Cursor SDK alters the way access is granted. The agent is no longer invoked by a developer at a keyboard. Instead, it can be called programmatically, either from a CI/CD trigger, a service on the backend, or directly embedded in another product.
Imagine this: Previously, it was necessary to be “in” To use the agents, you can click on its cursor. You can now call the same agents anywhere within your stack using a few TypeScript lines.
Starting is as simple as a command.
Then, create an Agent instance, send it a task, and stream the response back — all in TypeScript. Cursor announced a minimal example:
import { Agent } from "@cursor/sdk";
const agent = await Agent.create({
apiKey: process.env.CURSOR_API_KEY!,
model: { id: "composer-2" },
local: { cwd: process.cwd() },
});
const run = await agent.send("Summarize what this repository does");
Forewarn (const. of Run.Stream()) {
console.log(event);
}It is important to note that the word “you” means “you”. Agent.create() Accepts calls apiKeyA The model is a You can specify the model you want to use in either the field or by selecting one of two options: Local The following are some examples of how to use The cloud Configuration based on the location where execution is desired.
How to build your own agent stack
It’s important to understand the SDK and the problems it addresses before diving in. To build fast, reliable and capable agents for coding that can run against data safely, it takes a lot of engineering work: Secure sandboxing; durable state management and session setup; environment setup and context management. Dev teams are often forced to completely rewrite their agent loops to be able to use a newly released model. Cursor SDK removes the complexity of this infrastructure so that teams can concentrate on creating useful agents rather than maintaining it.
The Agent Harness – What is it? “Same Runtime” What Does It Mean?
SDK agents utilize the same Harness that Cursor uses to power its own products. ‘Harness’ here refers to the full set of supporting infrastructure that makes an agent effective beyond just the LLM call itself. Cursor is a good example. This includes:
Intelligent context management — Codebase indexing, semantic search, and instant grep so agents retrieve the right code context before generating responses. It is important because LLMs can only be as effective as the code context that they are given. Poor retrieval results in hallucinations or irrelevant outputs.
MCP Servers — Agents launched through the SDK can connect to external tools and data sources over Stdio You can access HTTP via any device. .cursor/mcp.json MCP (Model Context Protocol) is an open standard for wiring tools into agent runtimes. MCP is an open-standard for wiring tools to agent runtimes.
The following are some of the skills that you can learn: — Agents automatically pick up reusable behavior definitions from a .cursor/skills/ The directory is located in the repository.
Hooks — A .cursor/hooks.json file lets you observe, control, and extend the agent loop across cloud, self-hosted, and local runtimes — useful for logging, guardrails, or custom orchestration.
Subagents — The main agent can delegate subtasks to named subagents with their own prompts and models via the Agent Tool that allows multi-agent workflows to be created without the need for custom orchestration codes.
Cloud deployment: persistent, sandboxed and resumable
Cloud execution is one of the most useful features in the SDK. Cursor’s cloud allows you to configure each agent with its own VM that includes a strong sandbox, a cloned repository of your target, and a development environment fully configured. Critically, the agent keeps running even if the initiating machine goes offline — the developer can reconnect and stream the conversation later.
The cloud agents are integrated with Cursor’s Agents Window, web application and SDK. This allows a programmatically started task to be examined or manually taken over within the Cursor interface. When the agent finishes, it can open a PR, push a branch, or attach demos and screenshots — making them suitable for asynchronous, unattended workflows:
const agent = await Agent.create({
apiKey: process.env.CURSOR_API_KEY!,
model: { id: "gpt-5.5" },
cloud: {
repos: [{ url: "https://github.com/cursor/cookbook", startingRef: "main" }],
autoCreatePR: true,
},
});
const run = await agent.send("Fix the auth token expiry bug");
console.log(`Started ${run.id}`);
Check back later from any location:
const result = await (
await Agent.getRun(run.id, { runtime: "cloud", agentId: run.agentId })
).wait();
console.log(result.git?.branches[0]?.prUrl);
The SDK supports both self-hosted and hosted workers for dev teams that have security needs. Both code execution and the tool’s execution will remain within the network of the organisation.
Model Flexibility, Composer 2,
Cursor SDK provides access to all supported models. To switch between models, you only need to make a small change. The model is a The parameter allows for teams to assign tasks according to their cost and ability. Cursor’s Own Composer 2. — described as a specialized coding model achieving frontier-level performance at a fraction of the cost of general-purpose models — is positioned as the default recommendation for most coding agent tasks.
Getting Started
Cursor published a Public Adoption Guide to accelerate adoption. cookbook repository On GitHub, there are four projects to get you started: a quickstart that is a Node.js sample that creates a locally-installed agent, streams the response, a prototyping web tool that scaffolds new projects within a sandboxed environment, a kanban board powered by agents that opens PRs automatically when engineers move a card and a CLI that allows users to spawn Cursor Agents from their terminal.
The Cursor application has been released. Cursor SDK plugin Cursor Marketplace helps developers build directly within the editor.
What you need to know
- Public beta for Cursor SDK now available — Cursor has released a TypeScript SDK (
Install @cursor/sdk using npmThe Cursor Desktop App, the CLI and the web interface are powered by the Cursor Runtime. - The hard work of building code agents is eliminated — Teams no longer need to engineer secure sandboxing, durable state and session management, environment setup, and context management from scratch — and won’t need to rework agent loops every time a new model ships.
- Cursor can run locally on your computer or cloud — Agents can execute on a developer’s local machine for fast iteration, on Cursor’s cloud against a dedicated VM with strong sandboxing, or on self-hosted workers for teams with strict network security requirements.
- The full harness is included in the package — SDK agents inherit Cursor’s complete infrastructure: intelligent context management (codebase indexing, semantic search, instant grep), MCP server support, Skills, Hooks, You can also find out more about the following: Subagents — the same stack powering Cursor’s own products.
Check out the Cookbook and Technical details. Also, feel free to follow us on Twitter Don’t forget about our 130k+ ML SubReddit Subscribe now our Newsletter. Wait! What? now you can join us on telegram as well.
You can partner with us to promote your GitHub Repository OR Hugging Page OR New Product Launch OR Webinar, etc.? Connect with us
Michal Sutter, a data scientist with a master’s degree in data science from the University of Padova is an expert. Michal is a data scientist with a background in machine learning, statistical analysis and data engineering.
