Browser Mcp
@Agent360dk
Browser Mcp について
Drive your real, logged-in Chrome from any AI agent (Claude Code, Cursor, VS Code) — works where headless dies. Reads emailed login codes from your Gmail, solves CAPTCHAs, 34 tools. MIT, local-only.
基本情報
設定
以下の設定を使って、このサーバーを MCP 対応クライアントに追加してください。
{
"mcpServers": {
"browser-mcp": {
"command": "npx",
"args": [
"@agent360/browser-mcp"
]
}
}
}ツール
ツールは検出されませんでした
ツールは README から自動的に抽出されます。メンテナーは ## Tools という見出しの下に記載することで、このタブに反映できます。
概要
What is Browser MCP?
Browser MCP gives AI agents (Claude Code, Cursor, VS Code agent mode) control of your real, logged-in Chrome browser — with your cookies, sessions, and 2FA — so it works on CAPTCHA, 2FA, and anti-bot sites where headless tools get blocked. It is built for developers who need to automate web tasks that require an authenticated human browser.
How to use Browser MCP?
Install the MCP server by running npx @agent360/browser-mcp install, then load the Chrome extension from ~/.browser-mcp/extension/ via chrome://extensions with Developer mode enabled. Alternatively, install from the Chrome Web Store and run npx @agent360/browser-mcp install --skip-extension to configure Claude Code.
Key features of Browser MCP
- 34 browser control tools (navigation, interaction, tabs, data, CAPTCHA solving)
- ~80% reCAPTCHA checkbox auto-solve with human fallback
- Multi-session support with color-coded tab groups (up to 10 concurrent sessions)
- Human-in-the-loop overlay for 2FA, CAPTCHA, and credential input
- 9 built-in provider integrations for token extraction (Stripe, HubSpot, Slack, etc.)
- MIT licensed, free, and 100% local — nothing leaves your machine
Use cases of Browser MCP
- Operate web platforms that have no API (e.g., LinkedIn, Reddit, dashboards)
- QA your own web application end-to-end using a real browser session
- Automate workflows that require 2FA or CAPTCHA bypass by reusing your logged-in browser
- Extract API tokens from provider dashboards without manual copy-paste
FAQ from Browser MCP
How does Browser MCP differ from Playwright MCP?
Browser MCP uses your real Chrome with existing logins and cookies, supports multi-session with tab groups, includes human-in-the-loop tools, and has built-in provider integrations. Playwright MCP uses a headless browser that must log in every time and has no human fallback.
How does Browser MCP handle CAPTCHAs?
It auto-detects reCAPTCHA v2/v3, hCaptcha, Turnstile, and FunCaptcha. It can auto-click the checkbox (~80% success with Google login), use AI vision for grid selection, or fall back to asking the human via an overlay dialog.
Is Browser MCP free and does it send data anywhere?
Yes, it is MIT licensed and free. All processing is local — nothing leaves your machine. The extension connects to the MCP server via WebSocket on localhost.
How do I update Browser MCP?
The MCP server updates automatically via npx @agent360/browser-mcp@latest. The extension updates automatically if installed from the Chrome Web Store; if loaded unpacked, re-run npx @agent360/browser-mcp install and reload the extension in Chrome.
What are the runtime requirements?
You need Chrome (or a Chromium-based browser) and Node.js to run the npx command. No additional authentication is required — it uses your existing Chrome session.
「ブラウザ自動化」の他のコンテンツ
Firecrawl Mcp Server
mendableai🔥 Official Firecrawl MCP Server - Adds powerful web scraping and search to Cursor, Claude and any other LLM clients.
Fetch MCP Server
zcaceresA flexible HTTP fetching Model Context Protocol server.
Webpage Screenshot MCP Server
ananddtyagiAn MCP that gives your agent the ability to snap a screenshot of webpages. Useful when you want your agent to check its progress during development.
Fastcrw
usFast, lightweight Firecrawl/Tavily alternative in Rust. Web scraper, crawler & search API with MCP server for AI agents. Drop-in Firecrawl-compatible API (/scrape, /crawl, /search). 2.3x faster than Tavily, 1.5x faster than Firecrawl in 1K-URL benchmarks. 6 MB RAM, single binary.
Fetch MCP
jae-jaeMCP server for fetch web page content using Playwright headless browser.
コメント