RELATED 01
Browser Use
OFFICIAL SERVERLocal MCP server that drives a Chrome or Chromium browser step by step, with an agent fallback
120kGitHub stars
8.1M/moPyPI downloads
119popularity score
Browser Use Cloud MCP
Hosted endpoint that hands whole web tasks to Browser Use's cloud browser agent and returns the result
Official · browser-use.com Remote (hosted) · Streamable HTTP Browser automationAI and ML
Connect your client
Remote (hosted)
Add in one click
Opens the app, which asks before it adds the server. Then replace each <NAME> in its settings with your value.
Or add it by command or config
Claude CodeTerminal
claude mcp add --transport http browser-use-cloud https://api.browser-use.com/v3/mcp --header 'x-browser-use-api-key: <BROWSER_USE_API_KEY>'Cursor~/.cursor/mcp.json
{
"mcpServers": {
"browser-use-cloud": {
"url": "https://api.browser-use.com/v3/mcp",
"headers": {
"x-browser-use-api-key": "<BROWSER_USE_API_KEY>"
}
}
}
}VS Code.vscode/mcp.json
{
"servers": {
"browser-use-cloud": {
"type": "http",
"url": "https://api.browser-use.com/v3/mcp",
"headers": {
"x-browser-use-api-key": "<BROWSER_USE_API_KEY>"
}
}
}
}Codex~/.codex/config.toml
[mcp_servers.browser-use-cloud]
url = "https://api.browser-use.com/v3/mcp"
http_headers = { "x-browser-use-api-key" = "<BROWSER_USE_API_KEY>" }Gemini CLITerminal
gemini mcp add --scope user --transport http --header 'x-browser-use-api-key: <BROWSER_USE_API_KEY>' browser-use-cloud https://api.browser-use.com/v3/mcpWhat it does
Browser Use Cloud MCP is the remote MCP endpoint of Browser Use Cloud, the company's paid service for running browser agents. Instead of exposing clicks and keystrokes one by one, it takes a task written in plain language, starts a cloud browser session, and lets Browser Use's own agent carry the task out. The assistant then checks progress, reads the agent's steps and fetches the final output.
The seven tools cover the session lifecycle: run_session starts a session with a task (optionally kept alive, with a chosen model such as claude-sonnet-4.6, claude-opus-4.6 or gpt-5.4-mini, a JSON output_schema, or a saved browser profile_id); get_session reports status, step count, cost and a live view URL; send_task gives an idle kept-alive session a follow-up; stop_session ends either the current task or the whole sandbox; and the remaining tools list sessions, their messages and saved browser profiles.
The endpoint sits on the v3 API (https://api.browser-use.com/v3/mcp). The provider's own documentation index recommends API v4 for new agent integrations, so the MCP endpoint is not on the newest API generation (llms.txt). It is a separate product from the open-source browser-use Python library, which has its own API and runs locally.
What it needs
A Browser Use Cloud account and API key from cloud.browser-use.com/settings, sent as the x-browser-use-api-key header. Usage is pay as you go from prepaid credits; signup credits are a one-time grant. Any MCP client that supports remote streamable HTTP servers with custom headers; nothing to install locally.
Limits
Runs tasks only through Browser Use's own agent and models; there are no low-level click, type or screenshot tools. Each run costs credits. Concurrent sessions are capped per project by lifetime spend: 10 at $0, rising to 50, 250, 500 and 1000 at $100, $1000, $5000 and $25000 (concurrency guide). API-key monthly spending caps are soft, so running work can go past them. Uses the v3 API, while the docs point new integrations to v4.
Who it suits
Suits developers who want an assistant to delegate complete multi-step web jobs (form filling, research, extraction into a schema) without running a browser on their own machine, and who accept paying per task and sending page content through Browser Use's cloud. Those who need fine-grained page control, offline use or no third-party service should look at a local browser server instead, such as the open-source browser-use library's own MCP mode, which we also list and which runs on the user's machine.
Tools 7
run_session- Start a cloud browser session and run a task in it, with optional model, output schema, keep-alive and profile
get_session- Check a session's status, step count, cost and live view link
send_task- Give a follow-up task to an idle session that was kept alive
stop_session- Stop the running task only, or tear down the whole session sandbox
get_session_messages- Read the agent's browser actions, reasoning and results for a session
list_sessions- List recent sessions with their status and cost
list_browser_profiles- List saved browser profiles usable for logged-in tasks