Fetch
Read any page reliably
Pull clean content from any URL, with automatic fallback from lightweight HTTP to full browser rendering when the page requires it.
The most powerful CLI for your agents to browse the web.
One CLI for skills, browser primitives, debugging, and cloud sessions, designed to be driven by an agent.
Three core features
Search
Instant, high-quality search
Run structured web searches through a single command. Power through Browserbase Search when you need it. No extra API keys, no extra accounts.
Built on Stagehand
Stagehand gives Browse real browser control for your agents.
Browse is built on Stagehand, the AI SDK for browser agents. That gives your agents a real browser they can navigate, click, type into, scroll, inspect, and extract structured content from without brittle manual scripting.
Stagehand also powers the hard parts behind Browse: accessibility-tree awareness, multi-step auth flows, JavaScript-heavy apps, and the same workflows running locally in Chrome or remotely on Browserbase in production.
Manage the full Browserbase platform from your terminal.
The CLI goes beyond browsing. Manage sessions, persist state, deploy automation, and extend your browser — all without leaving the terminal.
Sessions and observability
Create, list, inspect, and debug remote browser sessions. View logs, watch session recordings, download artifacts, and manage the full session lifecycle with bb sessions.
Persistent contexts
Persist cookies, localStorage, and browser state across sessions with bb contexts. Your agent stays logged in without re-authenticating every run.
Serverless functions
Scaffold, develop locally, and deploy browser automation to the cloud with bb functions. Init, dev, publish, invoke — the full lifecycle, no infrastructure to manage.
One CLI. The whole browser workflow.
Install the CLI, set your API key, and start browsing. Works locally for development and routes through Browserbase in production.
npm install -g browse
Built for coding agents
Built for the way coding agents work.
Agent-ready by default. Browserbase CLI is terminal-native and predictable, so agents can inspect commands with --help, parse structured output, and chain browser actions without brittle glue code.
Skills for real workflows. Run bb skills to install reusable Browserbase workflows for your agent. The ecosystem at skills.sh/browserbase gives agents higher-level workflows for common browser tasks.
From local dev to production. Start with a local browser while developing, then move the same workflow onto Browserbase when you need hosted browsers, persistent state, observability, and reusable automation.
Frequently Asked Questions
What is Browserbase CLI used for?
Browserbase CLI gives agents one interface to fetch pages, search the web, control a real browser, manage sessions, persist browser state, and turn browser workflows into reusable automation. It is built for terminal-native agent workflows, so your agent can go from reading the web to acting on it without stitching together separate tools.
Does Browserbase CLI work with Claude Code, Cursor, and Codex?
Yes. Browserbase CLI is designed for coding agents: it is scriptable, discoverable, and works well inside terminal-based agent loops. In Claude Code, you can also run bb skills to install reusable Browserbase workflows. The same CLI fits naturally into Cursor and Codex-style workflows.
Can I use Browserbase CLI with a local browser?
Yes. You can use Browse with a local browser for development and debugging, then move the same workflow onto Browserbase when you need hosted browsers, better reliability, or production scale.
Install the CLI. Give your agent the web.
One command to fetch, search, browse, and manage the full Browserbase platform.
