steel-dev/steel-browser
Steel Browser
Steel is a cloud browser automation platform that provides a REST API for launching and controlling remote Chrome browser sessions. It enables programmatic browsing and web scraping using standard automation tools like Puppeteer, Playwright, and Selenium, connecting to cloud-hosted browser instances via WebSocket and the Chrome DevTools Protocol. The platform supports both headless and headful browser sessions, with language-specific SDKs for TypeScript and Python.
The service distinguishes itself through comprehensive anti-detection capabilities, including residential proxy rotation, CAPTCHA solving, browser fingerprint randomization, and human behavior simulation to evade bot detection systems. It maintains persistent browser state across sessions, preserving cookies, local storage, and authentication for multi-step workflows. Steel also offers natural-language browser automation, allowing AI agents to drive web interactions using plain-English instructions rather than low-level selectors.
Beyond core automation, the platform provides session monitoring and debugging tools with live streaming and recorded replays, file transfer capabilities, and content extraction features that capture screenshots, PDFs, and Markdown from fully rendered web pages. It supports mobile browser emulation, geographic traffic routing, and serverless execution from edge environments. The platform can be deployed as a self-hosted runtime using Docker, giving teams full control over the browser infrastructure.
Features
- Natural Language Browser Control Platforms - A platform that controls browsers using natural-language instructions and connects AI agents for autonomous web navigation.
- Cloud Browser Automation Scripts - A REST API that launches and manages remote Chrome sessions for programmatic browsing with Puppeteer, Playwright, or Selenium.
- Agentic Browser Controllers - Provides an API to connect AI agent frameworks for autonomous web navigation and interaction.
- Agentic Web Interaction - Controlling browsers using natural-language instructions and connecting AI agent frameworks for autonomous web interaction and data extraction.
- Autonomous AI Agent Frameworks - Integrates with AI agent frameworks to enable autonomous web task execution via browser control.
- Natural Language Command Translation - Translates plain-English instructions into executable browser automation commands using an AI agent framework.
- Natural Language Automation - Drives a browser using high-level instructions instead of selectors, translating plain-language commands into page actions.
- Web Page Scraping Extractors - Extracts clean, well-formatted HTML data from any webpage using a running server instance.
- JavaScript Rendering - Waits for single-page applications to fully render before scraping dynamically loaded data from the DOM.
- Browser Session Stream Viewers - Streams the remote browser's screen in real time via a web viewer for debugging live automation sessions.
- Browser Storage Persistence - Stores authentication, cookies, and configuration in reusable profiles so workflows maintain consistent identity across sessions.
- Browser Automation Framework Connectors - Controls remote browser sessions through Playwright, Puppeteer, and Selenium for existing automation workflows.
- Browser Automation Tool Connectors - Connects Puppeteer, Playwright, and Selenium scripts to managed cloud browser sessions.
- Natural Language Automation - Translates natural-language instructions into browser automation commands for AI-driven web interaction.
- Cloud Browser Provisioners - Creates a cloud-hosted Chrome instance on demand and exposes a CDP endpoint for remote control.
- Anti-Detection Session Launches - Creates a full Chrome instance in the cloud with anti-bot protection and returns a WebSocket URL for programmatic control.
- CDP WebSocket Tunnels - Exposes a remote Chrome DevTools Protocol endpoint over WebSocket for direct attachment of Puppeteer and Playwright clients.
- Proxy Traffic Management - Chains proxy servers for IP rotation and geolocation masking during automated browsing sessions.
- Anti-Detection Strategies - Activates stealth measures such as proxies and captcha solving on a cloud browser session to evade anti-bot systems.
- Anti-Bot Evasion - Activates proxy rotation, CAPTCHA solving, and session timeouts to bypass fingerprinting and challenge-based defenses.
- Fingerprint Randomization - Randomizes browser fingerprints, rotates residential proxies, and simulates human behavior to evade anti-bot detection systems.
- Human Behavior Simulations - Adds natural mouse movements and interaction patterns to avoid detection as an automated bot.
- Session-Based Authentication Proxies - Routes browser automation traffic through a proxy layer that injects authentication headers and manages session state.
- Stateful Session Persistence - Persists browser state across interactions so automation agents can resume work without losing context.
- Containerized Browser State Persisters - Persists browser state across requests within isolated containerized sessions for multi-step automation workflows.
- Browser Session Persistence - Saves and restores cookies, localStorage, and sessionStorage across sessions so automations resume without re-authenticating.
- Session-Context Debugging - Provides live viewing, session recordings, logs, and traces to debug flaky web automation workflows.
- Web Automation Session Debuggers - Provides live viewing, session recordings, logs, and traces to diagnose failures in flaky web automation workflows.
- Live and Recorded Session Viewers - Provides a viewer to inspect and debug browser sessions in real time or from a recording.
- Browser Control SDKs - Provides TypeScript and Python SDKs for programmatic browser control.
- Browser Fingerprint Managers - Controls proxy and fingerprint settings to avoid bot detection and maintain session anonymity.
- Automatic Provisioning and Cleanup - Manages provisioning, scheduling, and cleanup of browser instances automatically so teams avoid infrastructure overhead.
- Browser Session Persistence - Keeps browser sessions alive across multiple tasks to avoid repeated connection handshake overhead.
- JavaScript-Rendered Content Extractors - Waits for JavaScript-rendered content to fully load before extracting structured data from single-page applications.
- Remote Browser Controllers - Launches and manages headless Chrome sessions via a REST API, enabling programmatic browsing and automation.
- Browser Control APIs - Provides an HTTP-based API for programmatic Chrome control compatible with Playwright and Puppeteer.
- Browser Isolation Strategies - Creates isolated, cloud-hosted browser sessions that prevent state collisions between concurrent automation workflows.
- Browser Navigation - Opens fresh browser tabs and loads URLs, executing JavaScript and rendering dynamic content like a real user.
- Cloud Browser Integrations - Integrates Puppeteer, Playwright, and Selenium with cloud-hosted browser sessions via WebSocket.
- Remote Browser Connections - Connects Puppeteer and Playwright scripts to managed cloud browser instances via WebSocket.
- Browser Session Managers - Creates and controls isolated browser instances with custom proxies, dimensions, and ad blocking settings.
- Customizable Session Launches - Launches isolated Chrome instances with custom proxy, ad blocking, and viewport settings for stateful automation.
- Multi-Format Page Extractors - Scrapes HTML, converts pages to markdown or readability format, and captures screenshots or PDFs through simple API endpoints.
- Browser Automation WebSocket Bridges - Exposes WebSocket endpoints that bridge Puppeteer and Playwright to remote browser sessions.
- Simplified Page Layouts - Simplifies web interactions for AI agents by presenting linear, touch-optimized layouts with fewer elements.
- Agentic Workflow Automation - Integrates with agent-oriented SDKs and standard automation tooling to reduce glue code for AI-driven browser tasks.
- DOM Simplification for Token Reduction - Reduces AI token consumption by serving simpler DOM structures and faster-loading mobile pages.
- CAPTCHA Solving - Detects and solves reCAPTCHA challenges during automation, with status checking for ongoing CAPTCHA resolution.
- Web Page Markdown Converters - Converts a web page's readable content into clean Markdown format via a REST endpoint.
- URL Batch Processors - Processes multiple URLs concurrently using async concurrency controls to speed up batch browser automation tasks.
- Session File Transfers - Uploads files into cloud browser sessions and downloads individual files or entire archives from them.
- Cloud Browser File Transfers - Uploads, downloads, and organizes files within browser sessions for data exchange.
- DOM-Based Extractions - Runs JavaScript inside browser contexts to read DOM elements and return formatted data such as text and links.
- Browser Automation Scaffolders - Generates a local project with pre-configured Playwright and Steel session setup using a single CLI command, reducing initial configuration effort.
- Virtual Browser Hosting - Runs the browser engine on your own infrastructure using open-source code, giving full transparency and control over the environment.
- Open-Source Browser Runtimes - Deploys the open-source browser runtime on your own infrastructure using Docker or other common paths, giving full control and transparency over the browser layer.
- Browser Session Scalers - Automatically handles infrastructure scaling, session management, and performance optimization for any workload size.
- Web Page Screenshot and PDF Tools - Renders any URL as a PDF document via a dedicated API endpoint.
- Browser Device Emulation - Emulates mobile browsers by configuring viewport, touch events, and device fingerprints to load mobile-optimized websites.
- Session Timeout Configurations - Sets a custom duration for a cloud browser session so long-running automation tasks do not expire before they finish.
- Geographic Traffic Routing - Routes browser traffic through residential proxies in a specific country, state, or city to appear as a local user.
- Automated Captcha Solvers - Automates CAPTCHA solving through dedicated API endpoints to bypass visual challenges during browsing.
- Credential Storage - Safely persists and retrieves login credentials for reuse across automated browser sessions.
- Mouse Movement Simulations - Provides natural mouse movement simulation as a key anti-detection feature for automated browsing.
- Credential Storage - Saves and retrieves login credentials for automated authentication across browser sessions.
- Automated Website Login Agents - Securely signs agents into sites that require login, enabling automation behind authentication walls.
- Automation Error Types - Catches typed errors like rate limits and timeouts, enabling retry logic with exponential backoff for resilient scraping.
- Browser Session Streams - Provides live streaming of browser sessions for real-time monitoring of automation workflows.
- Browser Resource Blockers - Blocks ads, images, and stylesheets on request to reduce bandwidth and improve page-load performance.
- Browser Automation Retries - Re-attempts a browser automation task with exponential backoff and jitter when a session fails.
- Automation - Captures every navigation, click, and form fill during a session and replays them step by step in a dashboard.
- Automation Session Recorders - Captures action-by-action traces with live view and MP4 replay for debugging automated browsing flows.
- Wait Conditions - Blocks execution until a specific element or condition appears on the page, avoiding fixed delays.
- Edge Function Executions - Operates cloud browser sessions from edge functions like Vercel Edge or Cloudflare Workers without WebSocket management.
- Browser File Management - Uploads, downloads, and organizes files within the context of browser sessions.
- Multi-Step Orchestrators - Persists session state across form submissions to handle complex, multi-page workflows like applications or checkouts.
- Browser Extensions - Installs Chrome extensions into the managed browser instance to extend automation capabilities.
- Automatic Infrastructure Scalers - Manages browser installation, updates, and capacity automatically so users avoid manual infrastructure orchestration.
- Concurrent Browser Contexts - Increases the number of simultaneous browser sessions a plan supports for parallel agent execution.
- Session Timeout Configurations - Sets a custom timeout on a cloud browser session so long-running automation tasks do not expire before they finish.
- Element Wait Conditions - Pauses command execution until a specific element appears on the page, avoiding blind sleep calls.
- Retry and Backoff Logic - Re-attempts failed browser operations with exponential backoff and jitter to handle transient errors gracefully.
- Low-Latency Session Launchers - Creates remote browser instances and connects drivers in under a second to minimize cold-start latency.
- Puppeteer Connections - Attaches local Puppeteer clients to cloud browser sessions over WebSocket using API keys and session IDs.
- Headful Session Launchers - Launches visible browser windows to improve compatibility with sites that detect headless automation.
- Session Debugging Viewers - Provides live viewer URLs and event logs for debugging browser automation sessions.
- Full Page Screenshots - Takes full-page screenshots of web pages on demand through a simple API endpoint.
- LLM Development Frameworks - Browser sandbox and automation infrastructure for AI agents.
- AI Coding Assistants - Browser automation infrastructure for AI agents and development workflows.
