MCP Servers · browserbase

mcp-server-browserbase

Browserbase MCP Server is a TypeScript-based Model Context Protocol server that enables LLMs to control and interact with web browsers via Browserbase cloud automation and Stagehand AI capabilities. It provides six core tools for session management, navigation, page interaction, and data extraction with both hosted (SHTTP) and self-hosted (STDIO) deployment options.

Talk to DEV.co Discuss Implementation

Source: GitHub — github.com/browserbase/mcp-server-browserbase

3.4k

GitHub stars

358

Forks

TypeScript

Primary language

Apache-2.0

License (OSI-approved)

Key facts

Objective fields from the source. Values we can't verify are shown as “Unknown” rather than guessed.

Field	Value
Repository	browserbase/mcp-server-browserbase
Owner	browserbase
Primary language	TypeScript
License	Apache-2.0 — OSI-approved
Stars	3.4k
Forks	358
Open issues	51
Latest release	v3.0.0 (2026-03-31)
Last updated	2026-07-07
Source	https://github.com/browserbase/mcp-server-browserbase

What mcp-server-browserbase is

Self-hostable MCP server exposing Browserbase browser automation and Stagehand intelligent interaction through standardized tools (start, end, navigate, act, observe, extract). Built on TypeScript, supports multiple LLM backends (defaults to Gemini 2.5 Flash Lite, configurable to Claude/GPT-4o), deployable via NPM, Docker, or direct Node.js with STDIO/SHTTP transports.

Quickstart

Get the mcp-server-browserbase source

Clone the repository and explore it locally.

terminalbash

1git clone https://github.com/browserbase/mcp-server-browserbase.git2cd mcp-server-browserbase3# follow the project's README for install & configuration

Need it deployed, integrated, or customized instead? DEV.co ships production installs.

Best use cases

Autonomous Web Data Extraction & RPA

Use LLMs to navigate complex web UIs, observe dynamic content, and extract structured data without brittle selectors—ideal for price monitoring, job board scraping, form filling, and intelligent workflow automation.

AI-Powered Agent Workflows

Integrate with LLM applications (Claude, GPT-4) to build agents that research websites, interact with SaaS dashboards, and complete multi-step web tasks with natural language understanding.

LLM-Driven Testing & Validation

Leverage intelligent browser control for end-to-end testing, visual regression detection, and user flow validation where traditional selectors fail on dynamic or responsive interfaces.

Implementation considerations

Requires three sets of API credentials: Browserbase API key + project ID, and LLM provider key (Gemini by default, or custom OpenAI/Anthropic/etc. with --modelName and --modelApiKey).
Hosted (SHTTP) deployment recommended by maintainers for ease; self-hosted STDIO adds operational overhead but enables complete control and cost predictability.
Stagehand model selection impacts capability and cost—Gemini 2.5 Flash Lite is default; GPT-4o/Claude may perform better on complex tasks but incur higher LLM costs.
Session persistence is enabled by default (--persist true); configure --keepAlive, --contextId, and browser viewport dimensions (--browserWidth, --browserHeight) based on target application requirements.
Self-hosted setup requires Node.js build environment or Docker; ensure CI/CD integration if automating deployments across environments.

When to avoid it — and what to weigh

Strict Latency Requirements (< 500ms) — Cloud-based Browserbase sessions and LLM round-trips introduce multi-second latencies; not suitable for real-time, sub-second responsive applications.
Offline or Air-Gapped Environments — Requires internet connectivity to Browserbase cloud services and external LLM APIs (Gemini, OpenAI, Anthropic); cannot operate in fully isolated networks.
Heavy Browser Automation at Scale (1000+ concurrent sessions) — Cloud browser pooling and Browserbase API rate limits may constrain throughput; evaluate pricing and session limits for large-scale deployments.
Applications Requiring No Third-Party LLM Calls — Default and most capable configurations require external LLM APIs (Gemini, OpenAI, Anthropic) for Stagehand intelligence; cannot run purely offline without custom local models.

License & commercial use

Licensed under Apache License 2.0 (Apache-2.0), a permissive OSI-approved license that permits commercial use, modification, and distribution with proper attribution and liability disclaimer.

Apache 2.0 is permissive for commercial applications. No restrictions on commercial use in the license itself. However, costs arise from Browserbase cloud services and external LLM APIs (Gemini, OpenAI, Anthropic, etc.)—these are separate from the open-source license and depend on your service tier and consumption model. Review Browserbase and LLM provider commercial terms independently.

DEV.co evaluation signals

Editorial assessment — not user reviews. Directional, with an explicit confidence level.

Signal	Assessment
Maintenance	Active
Documentation	Strong
License clarity	Clear
Deployment complexity	Low
DEV.co fit	Strong
Assessment confidence	High

Security considerations

Browserbase cloud sessions inherit cloud provider security posture (unknown from this data). API key exposure risk: BROWSERBASE_API_KEY, BROWSERBASE_PROJECT_ID, and LLM API keys must be protected as secrets in self-hosted deployments. No claims made about encryption, audit logging, or data retention by this project itself. If handling sensitive websites or data, review Browserbase security documentation and your own network/access controls independently.

Alternatives to consider

Playwright / Puppeteer (direct)

Direct browser automation without LLM integration; lower latency and cost, but requires custom logic to parse and interact with pages; no intelligent observation/extraction.

Anthropic Computer Use (Claude API native)

Native LLM-to-desktop vision and control; operates on local/remote screens without explicit browser session management; different abstraction (not MCP protocol).

AgentQL + Other MCP Servers

Competing browser automation + semantic querying; some may offer alternative pricing, LLM backends, or MCP integration patterns; evaluate based on specific use case and cost.

Software development agency

Build on mcp-server-browserbase with DEV.co software developers

Browserbase MCP Server connects your LLMs to the web. Choose hosted (easiest) or self-hosted (full control). Start with free tier or contact us for enterprise deployment.

Talk to DEV.co

Related open-source tools

Surfaced by semantic similarity across the DEV.co open-source index.

mcp mobile-mcp servers

Related on DEV.co

Explore the category and the services that help you build with it.

MCP Servers AI Application Development Web App Development API Development Talk to DEV.co

mcp-server-browserbase FAQ

Can I use this with models other than Gemini?

Yes. Default is Gemini 2.5 Flash Lite. Use --modelName (e.g., anthropic/claude-sonnet-4.5) and --modelApiKey with your own LLM provider credentials. See docs for supported models in Stagehand.

Do I have to use the hosted Browserbase MCP server?

No. Hosted (SHTTP) is recommended for ease, but you can self-host via NPM or Docker with full control. Self-hosted adds operational overhead but reduces dependency on Browserbase-managed infrastructure.

What are typical costs?

Unknown from this data. Costs depend on Browserbase session usage (cloud browser provisioning) and LLM API consumption (Gemini/OpenAI/Anthropic). Hosted option may include LLM cost coverage; review Browserbase pricing page.

Is there a limit to concurrent browser sessions?

Not explicitly stated. Browserbase cloud automation and API rate limits will constrain concurrency. Check Browserbase plan documentation and API limits for your deployment scale.

Work with a software development agency

DEV.co helps companies turn open-source tools like mcp-server-browserbase into production software. Our software development services cover the full lifecycle — architecture, web development, integration, and maintenance — delivered by software developers and web developers who ship. Engage our software development agency to implement or customize it for your mcp servers stack.

Ready to Add Intelligent Web Automation to Your LLM App?

Browserbase MCP Server connects your LLMs to the web. Choose hosted (easiest) or self-hosted (full control). Start with free tier or contact us for enterprise deployment.

Talk to DEV.co Schedule a Consultation