MCP Clients (Claude Code, OpenClaw)

BrowserOS is the best browser for AI coding agents. It comes with a built-in MCP server that gives your AI agent full browser control and direct access to 40+ external services — Gmail, Slack, GitHub, Google Calendar, Linear, Notion, and more — all through a single MCP connection.

Unlike Chrome DevTools MCP which requires setting up debug profiles and running separate servers, BrowserOS MCP works out of the box. Just copy the URL from settings and connect.

Why Use BrowserOS with Claude Code?

Agentic Coding

Claude tests your web app, reads console errors, and fixes the code — all in one loop.

40+ App Integrations

Gmail, Slack, GitHub, Jira, Notion, Google Sheets, and more — accessible directly from your AI agent.

Data Extraction

Extract your LinkedIn profile, tweets, or any authenticated page content.

Task Automation

Fill forms, navigate multi-step workflows, and automate repetitive browser tasks.

53+ MCP Tools

Full browser control: tabs, navigation, clicks, typing, screenshots, bookmarks, history, tab groups, and window management.

Zero Config Auth

Connect external services via OAuth — credentials are managed securely, never stored in BrowserOS.

Wondering how BrowserOS MCP compares to Chrome DevTools MCP or other browser automation tools? See our detailed feature comparison covering 53 browser tools, 40+ app integrations, and why BrowserOS MCP gives developers more out of the box.

Getting Started

Open BrowserOS Settings

Navigate to chrome://browseros/mcp or click Settings → BrowserOS as MCP in the sidebar.

Copy the MCP URL

Copy the Server URL shown on the page (e.g., http://127.0.0.1:9239/mcp).

BrowserOS MCP settings page showing Server URL

Connect your MCP client

Use the tabs below to connect your preferred client.

Claude Code
Gemini CLI
Codex
OpenClaw
Claude Desktop

Add BrowserOS to Claude Code:

claude mcp add --transport http browseros <mcp_url> --scope user
# Example: claude mcp add --transport http browseros http://127.0.0.1:9239/mcp --scope user

Start Claude Code and try it:

claude
> Open amazon.com in BrowserOS

Run claude --dangerously-skip-permissions to skip confirmation prompts for each browser action.

To remove later:

claude mcp remove browseros --scope user

Add BrowserOS to Gemini CLI:

gemini mcp add local-server <mcp_url> --transport http --scope user
# Example: gemini mcp add local-server http://127.0.0.1:9239/mcp --transport http --scope user

Add BrowserOS to OpenAI Codex CLI:

codex mcp add browseros <mcp_url> --transport http
# Example: codex mcp add browseros http://127.0.0.1:9239/mcp --transport http

Add BrowserOS to OpenClaw by adding it to the mcpServers section in your openclaw.json config file:

{
  "mcpServers": {
    "browseros": {
      "url": "http://127.0.0.1:9239/mcp"
    }
  }
}

Replace the URL with your BrowserOS MCP URL from settings.

Open your Claude Desktop config file:

~/Library/Application Support/Claude/claude_desktop_config.json

Add BrowserOS:

{
  "mcpServers": {
    "browseros": {
      "command": "npx",
      "args": ["mcp-remote", "http://127.0.0.1:9239/mcp"]
    }
  }
}

Do not add BrowserOS to claude_desktop_config.json as a raw { "url": "..." } entry. Claude Desktop local config expects a command wrapper here, so use npx mcp-remote <mcp_url>.

Restart Claude Desktop to connect.

Example Prompts

Try these prompts once connected

Extract structured data from any page you’re logged into — no scraping setup needed.

Open my LinkedIn profile in BrowserOS and extract my work experience as JSON

Test your web app end-to-end without leaving the terminal — Claude navigates, interacts, and reports errors back.

In BrowserOS, go to localhost:3000, click the login button, and check for console errors

Capture visual snapshots of any page for debugging, documentation, or design review.

Take a screenshot of the current page in BrowserOS and save it to screenshots/

Search through your real browsing history to find pages you visited earlier.

Search my BrowserOS history for "invoice" and list the recent matches

Access your email directly from the agent — read, search, and summarize without switching windows.

Read my latest Gmail messages and summarize them

Chain multiple services together in a single prompt — file issues, notify your team, and stay in flow.

Create a Linear issue for the bug I just found and post a summary to Slack

Browser Automation Tools

BrowserOS exposes 53 browser automation tools to MCP clients, organized into the following categories:

Navigation & Tabs (8 tools)

Tool	Description
`get_active_page`	Get the currently focused page/tab
`list_pages`	List all open pages with title, URL, and tab ID
`navigate_page`	Navigate to a URL, or go back/forward/reload
`new_page`	Open a new tab (with optional background/hidden mode)
`new_hidden_page`	Open a hidden tab for background automation
`show_page`	Restore a hidden page to visible state
`move_page`	Move a tab to a different window or position
`close_page`	Close a tab

Content & Observation (8 tools)

Tool	Description
`take_snapshot`	Get accessibility tree with interactive element IDs
`take_enhanced_snapshot`	Detailed accessibility tree with structural context
`get_page_content`	Extract page as clean Markdown (headers, links, tables)
`get_page_links`	Extract all links from a page with deduplication
`get_dom`	Get raw HTML DOM with optional CSS selector scoping
`search_dom`	Search DOM by text, CSS selector, or XPath
`take_screenshot`	Capture page screenshot (PNG/JPEG/WebP, full-page option)
`evaluate_script`	Execute JavaScript in the page context

Interaction & Input (14 tools)

Tool	Description
`click`	Click an element by ID from snapshot
`click_at`	Click at specific X,Y coordinates
`hover`	Hover over an element
`focus`	Focus an element (scrolls into view)
`fill`	Type text into an input (with optional clear)
`clear`	Clear text from input/textarea
`check`	Check a checkbox or radio button
`uncheck`	Uncheck a checkbox
`select_option`	Select a dropdown option by value or text
`press_key`	Press a key or key combination (Enter, Ctrl+A, etc.)
`drag`	Drag an element to another element or coordinates
`scroll`	Scroll page or element (up/down/left/right)
`upload_file`	Upload files to a file input element
`handle_dialog`	Accept or dismiss JavaScript dialogs

File & Export (3 tools)

Tool	Description
`save_pdf`	Print current page to a PDF file
`save_screenshot`	Capture and save a screenshot to disk
`download_file`	Click an element to trigger a download and save it

Window Management (5 tools)

Tool	Description
`list_windows`	List all browser windows
`create_window`	Create a new browser window
`create_hidden_window`	Create a hidden window for background tasks
`close_window`	Close a window by ID
`activate_window`	Focus and activate a window

Tab Groups (5 tools)

Tool	Description
`list_tab_groups`	List all tab groups
`group_tabs`	Create a tab group with optional title and color
`update_tab_group`	Update group title, color, or collapsed state
`ungroup_tabs`	Remove tabs from groups
`close_tab_group`	Close a group and all its tabs

Bookmarks (6 tools)

Tool	Description
`get_bookmarks`	List all bookmarks
`create_bookmark`	Create a bookmark or folder
`remove_bookmark`	Delete a bookmark or folder
`update_bookmark`	Update bookmark title or URL
`move_bookmark`	Move a bookmark to a different folder
`search_bookmarks`	Search bookmarks by title or URL

History (4 tools)

Tool	Description
`search_history`	Search browser history by text query
`get_recent_history`	Get the most recent history items
`delete_history_url`	Delete a specific URL from history
`delete_history_range`	Delete history within a time range

40+ External App Integrations

BrowserOS connects your AI agent directly to the tools you already use — no separate MCP servers to install or configure. Everything is accessible through the same BrowserOS MCP connection.

How It Works

Agent calls an external service tool

Your AI agent calls a tool like gmail_search_messages through the BrowserOS MCP.

OAuth login (first time only)

If this is your first time using that service, BrowserOS opens an OAuth login page in the browser. Log in and authorize access.

Tool executes and returns results

Once authenticated, the tool runs and returns results to your agent. Future calls to the same service work automatically — no re-authentication needed.

Your credentials are managed securely via OAuth and are never stored in BrowserOS. Tokens are refreshed transparently, and you can revoke access at any time from the service provider.

Supported Services

Service	What you can do
Gmail	Send, read, search emails, manage drafts and labels
Outlook Mail	Send, read, and manage emails
Resend	Send transactional and marketing emails

Calendar & Scheduling

Service	What you can do
Google Calendar	Create events, find free time, manage calendars
Outlook Calendar	Schedule meetings, manage events
Cal.com	Schedule meetings, manage availability

Communication

Service	What you can do
Slack	Post messages, manage channels
Discord	Send messages, manage servers
WhatsApp	Send messages, manage conversations
Microsoft Teams	Chat, meet, and collaborate

Development

Service	What you can do
GitHub	Manage repos, issues, and pull requests
GitLab	Manage repos, issues, and merge requests
Vercel	Deploy and manage web applications
Postman	Test and manage APIs
Cloudflare	Manage domains, DNS, and security
Supabase	Manage databases and backend services

Project Management

Service	What you can do
Linear	Create issues, manage cycles and projects
Jira	Create issues, manage sprints
Asana	Organize and track team projects
Monday	Manage work and team collaboration
ClickUp	Manage tasks, projects, and workflows

Productivity & Docs

Service	What you can do
Notion	Create pages, manage databases
Google Docs	Create and edit documents
Google Sheets	Create and edit spreadsheets
Google Drive	Upload, download, and manage files
Google Forms	Create and manage forms and surveys
Confluence	Create and manage documentation
Airtable	Manage bases, tables, and records

File Storage

Service	What you can do
Dropbox	Store and share files
OneDrive	Store and sync files with Microsoft
Box	Manage and share enterprise files

Design

Service	What you can do
Figma	Access and manage design files
Canva	Create and manage designs

CRM & Marketing

Service	What you can do
Salesforce	Manage leads, contacts, and opportunities
HubSpot	Manage contacts, deals, and marketing

E-commerce & Payments

Service	What you can do
Shopify	Manage products, orders, and store
Stripe	Manage payments and subscriptions

Analytics

Service	What you can do
PostHog	Query analytics, manage feature flags
Mixpanel	Analyze user behavior and metrics

Support

Service	What you can do
Zendesk	Manage support tickets and customers
Intercom	Manage customer messaging and support

Search & AI

Service	What you can do
Brave Search	Search the web privately
Exa	AI-powered semantic web search
Mem0	Store and retrieve AI memory

Service	What you can do
LinkedIn	Post updates, manage connections
YouTube	Access video info and transcripts
WordPress	Manage websites and blog content

Demo Videos

Agentic coding workflow

Browser automation with Claude Code

Get Started

BrowserClaw

Core Features

Comparisons

Integrations

Troubleshooting

Contributing

Why Use BrowserOS with Claude Code?

Agentic Coding

40+ App Integrations

Data Extraction

Task Automation

53+ MCP Tools

Zero Config Auth

Getting Started

Example Prompts

Browser Automation Tools

40+ External App Integrations

How It Works

Supported Services

Demo Videos

Agentic coding workflow

Browser automation with Claude Code

​Why Use BrowserOS with Claude Code?

Agentic Coding

40+ App Integrations

Data Extraction

Task Automation

53+ MCP Tools

Zero Config Auth

​Getting Started

​Example Prompts

​Browser Automation Tools

​40+ External App Integrations

​How It Works

​Supported Services

​Demo Videos

​Agentic coding workflow

​Browser automation with Claude Code

Why Use BrowserOS with Claude Code?

Getting Started

Example Prompts

Browser Automation Tools

40+ External App Integrations

How It Works

Supported Services

Demo Videos

Agentic coding workflow

Browser automation with Claude Code