---
site: steel.dev
collection: docs
canonical_url: https://steel.dev/docs
markdown_url: https://steel.dev/docs.md
content_signal: ai-input=yes, ai-train=yes, search=yes
title: Steel
description: Steel gives AI agents real browsers in the cloud. Your agent uses websites the way a person does, and you can see everything it did.
summary: Steel gives AI agents real browsers in the cloud.
type: overview
reviewed: 2026-11-16
prerequisites: []
agent_setup: https://setup.steel.dev/install.md
docs_index: https://steel.dev/docs/sitemap.md
---

# Steel

Steel gives AI agents real browsers in the cloud. Your agent uses websites the way a person does, and you can see everything it did.

## Start in one step

Choose who drives the browser.

### Your coding agent

Open Claude Code, Codex, Cursor or another coding agent in your project, and send it this prompt:

```text
Read https://setup.steel.dev/install.md and follow it to set up Steel for this coding client.
Then continue my requested task using the installed Steel skill.
```

Approve the sign-in link that the agent gives you. Then give it a task, such as `Use Steel to open a browser, send me the live view link, and wait until I say I'm watching. Then open https://steel.dev/practice/news, list the top 3 stories, and stop the session.` [Set up your agent](https://steel.dev/docs/get-started/agent-setup) walks through each step, and has the setup for each agent.

### A chat agent

In Claude, go to **Customize > Connectors**, and add a custom connector with this server URL. Sign in to Steel when it asks. Any other client that supports remote MCP servers works the same way.

```text
https://mcp.steel.dev/mcp
```

[Connect Claude to Steel](https://steel.dev/docs/integrations/claude) has the steps. Another client adds the same URL in its own settings, as [MCP server](https://steel.dev/docs/control/mcp-server#connect-a-client) shows.

### Your own code

Install the SDK and Playwright with `npm install steel-sdk playwright` or `pip install steel-sdk playwright`, set `STEEL_API_KEY`, and run this program. It opens a page in a Steel session and prints its title.

**TypeScript**

```ts file=home.mts
import Steel from "steel-sdk";
import { chromium } from "playwright";

const client = new Steel();
const session = await client.sessions.create();

try {
  const browser = await chromium.connectOverCDP(session.websocketUrl);
  const context = browser.contexts()[0];
  const page = context.pages()[0] ?? (await context.newPage());
  await page.goto("https://steel.dev/practice/news");
  console.log(await page.title());
  await browser.close();
} finally {
  await client.sessions.release(session.id);
}
```

**Python**

```python file=home.py
from playwright.sync_api import sync_playwright
from steel import Steel

client = Steel()
session = client.sessions.create()

try:
    with sync_playwright() as playwright:
        browser = playwright.chromium.connect_over_cdp(session.websocket_url)
        context = browser.contexts[0]
        page = context.pages[0] if context.pages else context.new_page()
        page.goto("https://steel.dev/practice/news")
        print(page.title())
        browser.close()
finally:
    client.sessions.release(session.id)
```

[Write Steel code](https://steel.dev/docs/get-started/code) starts from this program, and sets up your coding agent to write Steel code for you. [Build your first Steel automation](https://steel.dev/docs/get-started/build-your-first-steel-automation) goes further: it signs in to a site, keeps the sign-in for the next run, and reviews the run.

Already use [Browser Use](https://steel.dev/docs/integrations/browser-use), [Playwright](https://steel.dev/docs/integrations/playwright) or [Puppeteer](https://steel.dev/docs/integrations/puppeteer)? Each guide runs a first program on Steel that you watch live, then moves the code that you have.

## Choose how to control a browser

| You want to | Use |
| - | - |
| Scrape a page as Markdown, or save it as a screenshot or a PDF, in one request | [Web scraping](https://steel.dev/docs/control/scrape) |
| Run Playwright or Puppeteer code that you already have | [CDP](https://steel.dev/docs/control/cdp) |
| Build an agent on a framework such as Browser Use, Stagehand or the OpenAI Agents SDK | [Integrations](https://steel.dev/docs/integrations) |
| Let a model read the page and act step by step | [Commands](https://steel.dev/docs/control/commands) |
| Drive the screen with a computer-use model | [Computer actions](https://steel.dev/docs/control/computer-actions) |
| Give a browser to an agent that runs shell commands, such as Claude Code, Codex or Cursor, or run browser steps from a terminal | [Steel CLI](https://steel.dev/docs/control/cli) |
| Give a browser to an agent that cannot run shell commands, such as Claude on the web | [MCP server](https://steel.dev/docs/control/mcp-server) |

[Control interfaces](https://steel.dev/docs/control) compares them in full.

## Popular

- [Get past bot detection](https://steel.dev/docs/stealth/get-past-bot-detection): Choose a stealth setup that fits how the site blocks your session.
- [Stay signed in across sessions](https://steel.dev/docs/logins/stay-signed-in): Keep an agent signed in from one session to the next with a saved profile.
- [Embed a live browser](https://steel.dev/docs/observability/embed-a-live-browser): Show a running session inside your product, interactive or view-only.
- [Debug a failed session](https://steel.dev/docs/observability/debug-a-failed-session): Read a session's signals, recording, logs and traces to find out why it failed.
- [Proxies](https://steel.dev/docs/stealth/proxies): Route a session through Steel's proxies, or bring your own.
- [API](https://steel.dev/docs/reference/api): Look up every endpoint and the SDKs.

## Get help

- [Troubleshooting](https://steel.dev/docs/troubleshooting): Find the error or problem that you see, and go to its fix.
- [FAQ](https://steel.dev/docs/faq): Read short answers to common questions.
- [Support](https://steel.dev/docs/support): Contact the Steel team.
