Skip to content

Redesign quickstart with agent-first one-command approach #278

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
mintlify wants to merge 1 commit into
base: main
Choose a base branch
from
Open

Redesign quickstart with agent-first one-command approach #278

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
122 changes: 71 additions & 51 deletions quickstart.mdx
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,30 +1,44 @@
---
title: "Quickstart"
description: "Get up and running with Kernel in one command."
---

import { YouTubeVideo } from '/snippets/youtube-video.mdx';

<Info>
If you are already familiar with browser vendors and are looking to quickly switch to Kernel browsers, follow the instructions **[here](/browsers/create-a-browser)**.
</Info>
## The fastest way to start

Paste this into your AI coding agent (Claude Code, Cursor, Windsurf, Goose, etc.) and let it handle the rest:

```
Install the Kernel CLI (brew install onkernel/tap/kernel), run kernel login, then kernel create to scaffold an app, and kernel deploy to deploy it. Use the Kernel docs at https://docs.kernel.sh for reference.
```

## Getting started
That's it. Your agent will install the CLI, authenticate you, scaffold a project from a template, and deploy it — all in one flow.

This quickstart guide will help you deploy and invoke your first browser automation on Kernel. You'll create a simple automation using Playwright, Computer Use, or a web agent framework like Browser Use.
<Tip>
For an even deeper integration, install the [Kernel MCP server](/reference/mcp-server) to give your agent direct access to Kernel's browser tools: `kernel mcp install --target cursor` (or `claude-code`, `windsurf`, `vscode`, `goose`, `zed`).
</Tip>

<YouTubeVideo videoId="trMR8HM0Zgs" title="Kernel Quickstart Video" />

## Prerequisites
---

## Manual setup

- `brew` for the Kernel CLI
- A [Kernel account](https://dashboard.onkernel.com/sign-up)
If you prefer to set things up yourself, follow these steps.

> **Note:** You can also deploy and invoke apps using the [Kernel MCP server](/reference/mcp-server) from AI assistants (Cursor, Goose, Claude, etc.).
<Info>
Already familiar with browser vendors? Skip to [creating a browser directly](/browsers/create-a-browser).
</Info>

### Prerequisites

- [Kernel account](https://dashboard.onkernel.com/sign-up)

## 1. Install the Kernel CLI
### 1. Install the Kernel CLI

```bash
# Using brew
# Using brew (recommended)
brew install onkernel/tap/kernel

# Using pnpm
Expand All @@ -34,50 +48,59 @@ pnpm install -g @onkernel/cli
npm install -g @onkernel/cli
```

Verify the installation exists:
Verify the installation:

```bash
which kernel
```

## 2. Create a new Kernel app
### 2. Authenticate

```bash
kernel create
kernel login
```

## 3. Authenticate with Kernel
This opens your browser to complete the OAuth flow. Credentials are stored and refreshed automatically.

The easiest way to authenticate is using OAuth:
### 3. Create an app

```bash
kernel login
kernel create
```

This will open your browser to complete the authentication flow. Your credentials will be securely stored and automatically refreshed.
The CLI walks you through picking a language (TypeScript or Python) and a template. To skip the prompts:

## 4. Deploy the sample app on Kernel
```bash
kernel create --name my-app --language ts --template sample-app
```

### 4. Deploy

<CodeGroup>
```bash Typescript / Javascript
cd sample-app
kernel deploy index.ts # --env-file .env if environment variables are needed
cd my-app
kernel deploy index.ts
```

```bash Python
cd sample-app
kernel deploy main.py # --env-file .env if environment variables are needed
cd my-app
kernel deploy main.py
```

</CodeGroup>

## 5. Invoke the app
Pass environment variables with `--env-file .env` if your template needs API keys.

### 5. Invoke

```bash
kernel invoke my-app get-page-title --payload '{"url": "https://www.google.com"}'
```

<AccordionGroup>
<Accordion title="More invoke examples by template">

<CodeGroup>
```bash Typescript / Javascript
# Sample app
kernel invoke ts-basic get-page-title --payload '{"url": "https://www.google.com"}'

# CAPTCHA Solver
kernel invoke ts-captcha-solver test-captcha-solver

Expand All @@ -98,9 +121,6 @@ kernel invoke ts-gemini-cua gemini-cua-task --payload '{"startingUrl": "https://
```

```bash Python
# Sample app
kernel invoke python-basic get-page-title --payload '{"url": "https://www.google.com"}'

# CAPTCHA Solver
kernel invoke python-captcha-solver test-captcha-solver

Expand All @@ -116,30 +136,30 @@ kernel invoke python-openai-cua cua-task --payload '{"task": "Go to https://news
# OpenAGI Computer Use
kernel invoke python-openagi-cua openagi-default-task --payload '{"instruction": "Navigate to https://agiopen.org and click the What is Computer Use? button", "record_replay": "True"}'
```

</CodeGroup>

## Next steps

Nice work! With Kernel, you:
</Accordion>
</AccordionGroup>

1. Developed an app that uses Playwright, Computer Use, or a web agent framework like Browser Use
2. Deployed and invoked it in the cloud
## Next steps

You can now update your browser automation with your own logic and deploy it again. Install our [MCP server](/reference/mcp-server) to give your coding agent our `search_docs` tool.
- Install the [MCP server](/reference/mcp-server) to give AI agents direct access to Kernel browsers
- Learn about [browser creation](/browsers/create-a-browser) for SDK-level control
- Explore [integrations](/integrations/overview) with Browser Use, Stagehand, Anthropic CUA, and more
- Browse available [sample apps](#sample-apps-reference) below

## Sample apps reference

These are the sample apps currently available when you run `kernel create`:

| Template | Description | Framework |
|-------------------------------|-----------------------------------------------------------|----------------------------|
| **sample-app** | Implements a basic Kernel app | Playwright |
| **captcha-solver** | Demo of Kernel's auto-CAPTCHA solving capability | Playwright |
| **browser-use** | Implements Browser Use SDK | Browser Use |
| **stagehand** | Implements the Stagehand v3 SDK | Stagehand |
| **anthropic-computer-use** | Implements an Anthropic computer use agent | Anthropic Computer Use API |
| **openai-computer-use** | Implements an OpenAI computer use agent | OpenAI Computer Use API |
| **gemini-computer-use** | Implements a Gemini computer use agent | Gemini Computer Use API |
| **openagi-computer-use** | Implements an OpenAGI computer use agent | OpenAGI Computer Use API |
| **magnitude** | Implements the Magnitude.run SDK | Magnitude.run |
These are the templates available when you run `kernel create`:

| Template | Description | Framework |
|----------------------------|--------------------------------------------------|----------------------------|
| **sample-app** | Basic Kernel app with Playwright | Playwright |
| **captcha-solver** | Demo of Kernel's auto-CAPTCHA solving | Playwright |
| **browser-use** | AI-powered web automation with Browser Use | Browser Use |
| **stagehand** | Stagehand v3 SDK integration | Stagehand |
| **anthropic-computer-use** | Anthropic computer use agent | Anthropic Computer Use API |
| **openai-computer-use** | OpenAI computer use agent | OpenAI Computer Use API |
| **gemini-computer-use** | Gemini computer use agent | Gemini Computer Use API |
| **openagi-computer-use** | OpenAGI computer use agent | OpenAGI Computer Use API |
| **magnitude** | Magnitude.run SDK integration | Magnitude.run |