We just launched Firecrawl Browser Sandbox - secure, fully managed environments where AI agents can interact with the web and extract data.

Firecrawl Browser Sandbox is a browser environment your agents can control - no local setup, no infrastructure to maintain.

• Zero config - no Chromium to install, no framework to configure
• Skill + CLI first - one command and your agent has a browser ready
• Fully managed & scalable - runs in isolated, disposable environments; launch hundreds of parallel sessions without managing anything
• Pre-loaded with Agent Browser - everything agents need to interact with the web, out of the box
• Works with Claude Code, Codex, OpenCode, and more

One call and your agent has a complete toolkit for handling any web task.

Firecrawl already gives your agents fast scrape and search endpoints. For most web tasks, that's enough.

But when it comes to pagination, form fills, and auth flows, agents need an integrated browser to navigate and extract hard-to-reach data. A static scrape won't get you through a login page. Search won't click through a multi-step pricing flow. And local browser tooling means CPU spikes, Chromium installs, and infrastructure you have to maintain.

This is why we're launching Firecrawl Browser Sandbox.

Browser Sandbox is a secure, remote browser environment your agents can control. Each session runs in an isolated container - fully managed by Firecrawl, with nothing running on your local machine.

Every sandbox comes pre-loaded with the tools agents already use:

• Playwright - for precise browser control and scripting
• Agent Browser - for autonomous, AI-driven navigation

No setup. No configuration. Call it from the CLI or API and your agent has a browser ready in seconds. Read the docs →

# Install CLI, auth, and add skills to all your agents
npx -y firecrawl-cli@latest init --all --browser
 
# Install Browser Sandbox via CLI
npm install -g firecrawl-cli
 
# Use Browser Sandbox
firecrawl browser "open uspto.gov"

from firecrawl import Firecrawl
 
app = Firecrawl(api_key="fc-YOUR-API-KEY")
 
response = app.list_browsers()
print(response.sessions)

Every session runs in an isolated, disposable environment. Agents and the sites they visit are fully contained within the sandbox, so you can run code securely without risk to your own infrastructure.

Since nothing runs locally, there are no CPU or RAM spikes on your machine. And you can launch parallel sessions - each in its own sandbox - scaling from one browser to hundreds without managing anything.

Browser Sandbox comes pre-loaded with Playwright and Agent Browser, so your agents have everything they need to interact with the web. No Chromium to install, no browser framework to configure.

Compatible with Claude Code, Codex, OpenCode, and more.

Browser Sandbox works alongside Firecrawl's existing scrape and search endpoints. Instead of stitching together separate tools, agents can navigate pagination, fill out forms, and work through multi-step flows to get data that scraping alone can't reach - all through a single platform.

When to use Browser and other Firecrawl endpoints:

Here's Firecrawl Browser Sandbox fetching dozens of patents with a single prompt:

Competitive intelligence

Your agent browses competitor sites, navigates search forms and filters, and extracts pricing, features, and positioning into structured data. Browser handles the interactive navigation; your agent handles the logic.

Knowledge base ingestion

Agents use Browser to navigate help centers, docs sites, and support portals that require clicks, pagination, and authentication - then pipe the extracted content into RAG pipelines to power AI support agents.

Market research at scale

Launch a swarm of agents, each with its own browser session, to build structured datasets from hundreds of pages - legal databases, job boards, real estate listings - in parallel.

Browser Sandbox is live:

• API - Full control in your applications
• CLI - One command to get started
• Playground - Try it immediately

Let your agents scrape, search, and browse. Securely.

Resources:

• Documentation: docs.firecrawl.dev/features/browser
• Get an API key: firecrawl.dev