← Back to Discover

Steel Browser

Open-source browser API and sandbox for AI agents that manages Chrome sessions, proxies and extensions, with scrape, screenshot and PDF endpoints.

Browser automation
Visit website ↗

What it does

Steel Browser is an open-source server that runs and manages Chrome for AI agents and web apps, so teams do not have to build their own browser infrastructure. It tracks sessions, pages and browser processes, and keeps cookies and local storage between requests. Agents can connect with Puppeteer, Playwright or Selenium, or call a REST API through Node and Python SDKs. Quick-action endpoints return a page as scraped content, a screenshot or a PDF, and the project also describes conversion to markdown and readable text. It includes proxy chain management for IP rotation, stealth plugins and fingerprint management, support for custom Chrome extensions, and a web UI for viewing and debugging sessions. You can run it in Docker, use one-click deploys to Railway or Render, or sign up for the hosted Steel Cloud. The project describes itself as being in public beta.

Capabilities

  • Session management that keeps cookies and local storage across requests
  • Control from Puppeteer, Playwright or Selenium
  • Scrape, screenshot and PDF endpoints
  • Proxy chains, stealth plugins and fingerprint management
  • Web UI and request logging for debugging sessions
  • Node and Python SDKs

When to use it

  • Giving an AI agent a controllable browser it can drive
  • Running browser automation on your own infrastructure
  • Pulling page content, screenshots or PDFs through an API