Guide · Open source
How to use claude-quickstarts: clone, run and adapt Claude API examples
Updated · written with AI from the project's own documentation and checked against our catalogue data
claude-quickstarts is a collection of example projects from Anthropic that show how to build applications on the Claude API. Each project is a starting point you can run, read and then adapt. This guide explains what you need before starting, how to get the code, and how to choose the right quickstart for your goal.
claude-quickstartsanthropics/claude-quickstarts · ★ 19.1k · MIT Install options & downloads Quick start
- 1Get a Claude API key. Sign up at console.anthropic.com. The README says you need a key to use these quickstarts.
- 2Install Git. Git is required to copy the repository. Install it if you do not already have it.
- 3Clone the repository. Run git clone https://github.com/anthropics/claude-quickstarts in the folder where you keep projects.
- 4Pick a quickstart. Choose the project that matches your goal, such as customer support, browser automation or a Managed Agents integration.
- 5Read the project README. Open the folder for your chosen quickstart and follow its setup steps, which vary by project.
- 6Adapt it to your needs. Change the code to fit your use case, and treat it as a starting point rather than production software.
What you need before you start
The README says you need a Claude API key. You can sign up for one at console.anthropic.com. You also need Git to copy the repository to your machine.
These quickstarts assume you are comfortable with code and configuration. They are not a ready-made app you can use without setup.
- A Claude API key from console.anthropic.com
- Git installed on your computer
- Comfort reading and configuring code
Get the code with git clone
Clone the repository to your machine with Git. Then open the folder for the quickstart you want to try. Each project lives in its own directory and links to its own page in the README.
git clone https://github.com/anthropics/claude-quickstartsChoose the quickstart that fits your goal
The collection covers several kinds of examples. Choose based on what you want to learn or build.
Customer Support Agent and Financial Data Analyst are good starting points for common business uses. Computer Use Demo, Computer Use Best Practices, Browser Use Demo and the Browser and Computer Toolset quickstarts cover automation. Autonomous Coding Agent uses the Claude Agent SDK. The Managed Agents projects show integrations with Slack, Linear, Sentry, Next.js and an MCP server.
- Customer Support Agent: AI-assisted support with access to a knowledge base
- Financial Data Analyst: analyze financial data via chat with interactive visualization
- Browser Use Demo: navigate sites, fill forms and extract content with a Playwright-backed tool
- Autonomous Coding Agent: a two-agent pattern that builds applications over multiple sessions
- Managed Agents: Linear, Slack, Sentry and Sentry Autofix: example integrations to study
- Managed Agents: MCP Server (TypeScript): lets Claude Desktop, Claude Code or claude.ai drive your workspace agents
Read the project before you change it
Each quickstart is meant to be read, run and then adapted. Several projects say they are reference or teaching examples rather than production code. The Browser Toolset and Computer Toolset quickstarts state that they show the interface and are not production code.
Some projects need extra setup. The Managed Agents Daily Brief uses ant apply to deploy files, and the Sentry Autofix project runs setup in Claude Code with Sentry's Agent Plugin. Follow the README inside each project folder for its exact steps.
- Check the project README for its own setup commands
- Look for notes saying a project is reference or teaching code
- Do not put production secrets into example code; the projects describe vaults for credentials
Run it safely
Several projects control a computer or browser. The Computer Use Best Practices README says this native-macOS project runs directly against the desktop and should be run in a VM. Use a sandbox or virtual machine for computer and browser automation examples.
Managed Agents projects can run on a schedule or react to webhooks. Read how each one is triggered before leaving it running.
- Run computer-use projects in a VM or isolated environment
- Review scheduled and webhook-driven projects before deploying them
What people use it for
- Learning how a customer support agent combines Claude with a knowledge base
- Evaluating whether Claude fits a data analysis task using the Financial Data Analyst example
- Studying a Slack, Linear or Sentry integration before building your own
- Understanding how an AI agent is wired together with tools and sessions
- Exploring browser automation with a Playwright-backed tool
Frequently asked questions
Is claude-quickstarts free?
The repository is licensed under MIT, which the listing states. The README says you can sign up for a Claude API key for free at console.anthropic.com, but API usage may be billed, and the source does not give pricing details.
What do I need to run claude-quickstarts?
You need a Claude API key from console.anthropic.com and Git to clone the repository. The README assumes you can set up and configure code.
How do I install claude-quickstarts?
Clone the repository with git clone https://github.com/anthropics/claude-quickstarts, then follow the README in the project folder you want to use. There is no single install step for the whole collection.
Can I use claude-quickstarts in production?
Not without review. Several projects say they are reference or teaching examples rather than production code, and the Browser and Computer Toolset quickstarts say they are not production code.
What are the alternatives to claude-quickstarts?
Comparable projects include LangChain, LlamaIndex, Vercel AI SDK and the OpenAI Cookbook. Those are different libraries or example collections, so pick based on whether you want Claude-specific examples or a general framework.
Why is my quickstart not working?
Check that your Claude API key is set up correctly and that you followed the README in that specific project folder. The source does not list specific troubleshooting steps, so consult the project README and its issues for errors.
New agent skills, repos and AI tools — every week
The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.
Our verdict
claude-quickstarts suits developers who want working Claude API examples to read and adapt, and students learning how agents use tools and sessions. It is a poor fit for non-developers wanting a finished app, or for anyone who needs production-ready code without review.
Install options & downloads