cursor/cookbook

★ 3,756TypeScriptAudience · developerComplexity · 3/5Setup · easy

Mindmap

mindmap
  root((Cursor Cookbook))
    What it does
      SDK usage examples
      Agent automation
      TypeScript recipes
    Examples
      Quickstart script
      App builder UI
      Agent kanban board
      CLI agent launcher
      DAG task runner
    SDK Features
      Stream responses
      Cloud agents
      Cancel and manage
    Setup
      Node.js required
      Cursor API key
      npm run commands

mindmap root((Cursor Cookbook)) What it does SDK usage examples Agent automation TypeScript recipes Examples Quickstart script App builder UI Agent kanban board CLI agent launcher DAG task runner SDK Features Stream responses Cloud agents Cancel and manage Setup Node.js required Cursor API key npm run commands

Click or tap to explore — scroll the page freely

Things people build with this

USE CASE 1

Run a Cursor coding agent from a terminal prompt using the CLI example without opening the editor UI

USE CASE 2

Build a web dashboard that lets you spin up multiple cloud Cursor agents and track their progress on a visual board

USE CASE 3

Break a large coding task into a dependency graph and run each piece in parallel with separate Cursor agents

USE CASE 4

Scaffold a new project automatically by sending a prompt to a Cursor cloud agent via the app-builder example

Tech stack

TypeScriptNode.jsCursor SDK

Getting it running

Difficulty · easy Time to first run · 5min

Requires a Cursor API key generated from the Cursor integrations dashboard.

License terms are not described in the README.

In plain English

Cursor Cookbook is a collection of small, ready-to-run examples for building software on top of Cursor, the AI-powered code editor. The examples are organized around the Cursor SDK, which is a TypeScript library that lets developers control Cursor's built-in coding agent from their own applications, scripts, and automated workflows. The SDK allows you to start an agent, send it a prompt, and receive a stream of responses as the agent works through a task. You can run agents locally against files on your machine or in the cloud, and the SDK gives you control over things like which AI model to use, how to cancel a running job, and how to manage the conversation history. The cookbook includes five example projects that show different ways to use the SDK. The quickstart is a minimal Node.js script that creates an agent, sends one prompt, and prints the response. The app-builder is a web interface for spinning up agents that scaffold new projects in an isolated cloud environment. The agent-kanban is a visual board for tracking multiple cloud agents, grouped by status or repository. The coding-agent-cli lets you start Cursor agents directly from a terminal prompt. The DAG task runner breaks a larger task into a dependency graph, runs each piece with a separate agent in parallel, and displays live status updates in Cursor's canvas view. To use any of the examples, you need a Cursor API key, which you can generate from the Cursor integrations dashboard. Once the key is set as an environment variable, each example can be run with standard Node.js commands. The full SDK documentation is available on the Cursor website.

Copy-paste prompts

Prompt 1

Using the cursor/cookbook quickstart, show me the minimal Node.js code to create a Cursor agent, send it a prompt to build a REST API, and stream the response to my terminal.

Prompt 2

Help me adapt the agent-kanban example from cursor/cookbook to track multiple agents working on different GitHub repos simultaneously.

Prompt 3

Show me how to use the DAG task runner from cursor/cookbook to split a large feature request into parallel subtasks and display live status in Cursor canvas.

Prompt 4

I have a Cursor API key, walk me through running the coding-agent-cli example from cursor/cookbook to start an agent from my terminal.

Open on GitHub → Explain another repo

← cursor on gitmyhub — every repo by this author, as a profile.

Verify against the repo before relying on details.