MCP Browser

You are doing a robot's job by hand

The answers are out there. People say exactly what they hate, want and would pay for. Getting to it is the slow part.

• Finding what people hate about a competitor

  Twenty Reddit tabs, a notes doc, and an afternoon you meant to spend building.

• Getting your AI to read a thread

  It says it cannot open the link, so you copy and paste the whole page into the chat. Again.

• Checking how your own site is doing

  Search Console in one tab, Analytics in another, Clarity in a third, and you still do the comparing yourself.

Ask it the way you would ask a person

No search operators, no scraping setup. Type the question in Claude or Cursor.

Founders"What do people wish [product category] tools did better?"

The top complaints, grouped by theme, with links to each thread.

Marketers"Which posts about [topic] got the most engagement on Facebook and X?"

The posts ranked by reactions, comments and shares, with the hooks they used.

Sales"Who is complaining about [competitor] on Reddit this month?"

The threads ranked by frustration, with a link to each person.

Recruiters"Which engineers sound frustrated with their job on LinkedIn?"

A shortlist of people worth messaging today.

Site owners"Which pages lost clicks this month, and what changed on them?"

Your real Search Console and Analytics numbers, compared for you.

Where it reads

Every one of these is a place your AI cannot reach on its own. With McpBrowser it reads them as you, because it uses the browser you are already signed in to.

The places your market talks

• Reddit MCP what people honestly think of your competitor
• X / Twitter MCP the trend three weeks before it peaks
• LinkedIn MCP who just switched jobs and needs what you sell
• Facebook MCP which posts your market is reacting to
• Pinterest MCP what shoppers will want next season
• Medium MCP how the experts are framing your space
• BlackHatWorld MCP the tactics nobody posts in public
• Dev.to MCP what developers are actually adopting
• Google News who is in the headlines today
• Google Trends what is rising and what is dying

Your own numbers

• Google Search Console MCP what your site really ranks for
• Google Analytics MCP which pages and channels actually work
• Microsoft Clarity where visitors get stuck
• Facebook page insights how your own page is doing

No API keys and no Google Cloud setup. If you can open the dashboard in Chrome, your AI can read it.

The three things people worry about

• Will my account get banned?

  It reads through your own Chrome session from your own Mac, and it only reads. It never posts, likes, follows or messages.

• Where do my logins go?

  Nowhere. Your cookies stay in Chrome on your Mac and are never sent to us.

• Do I need to be technical?

  No. It is a Mac app. There is no config file to edit and no terminal.

Running in about four minutes

1. Download the Mac app and open it.
2. Sign in with a free Webmatrices account.
3. Press Connect next to your AI app. It works with Claude Desktop, Claude Code, Cursor, Windsurf, Zed, VS Code. Other ways to install

Homebrew: brew install --cask bishwas-py/tap/mcpbrowser copy

Claude Desktop only: download the extension (.mcpb) and double-click it.

Free to start. One payment if you want more.

Free

$0

• 50 requests per day
• All 13 sources
• All 43 tools
• No credit card required Download for Mac, free

Unlimited

$9.99 once

• Unlimited requests
• All 13 sources
• All 43 tools
• No subscription, no renewal

Questions

My AI already searches the web. What does this give me that it cannot?

Web search sees the public, SEO-cleaned surface of the internet. McpBrowser uses your logged-in Chrome session, so it reaches the content that actually matters: the complaint thread in a login-required subreddit, the LinkedIn post your prospect liked but never shared publicly, the niche forum that only loads after sign-in. The gap isn't between AI tools. It's between what's public and what's behind a login.

Could Reddit or LinkedIn ban my account for using this?

They would have to ban your own browser. McpBrowser runs from your Mac with your real cookies, your normal IP, and a real Chromium fingerprint. Sites see a session that looks identical to you logging in and reading yourself, because that is exactly what it is. We never post, follow, like, or DM. Read-only research, nothing else.

What data leaves my Mac?

Your cookies never leave your Mac. Your logins stay inside your own Chrome and are never sent to us. The browsing itself runs locally too. When your AI uses a tool, McpBrowser saves a private record of that call to your Webmatrices account: which tool ran, what it asked for, and the text that came back. Passwords, tokens, and card details your AI passes to a tool are blanked out before anything is sent, and images are never uploaded. Only you and Webmatrices staff can see these records, and we never sell or share them.

I'm not technical. How long does setup actually take?

It is a Mac app. Download it, sign in with your Webmatrices account, press Connect next to your AI client. McpBrowser writes the config file for you. No JSON to edit, no terminal commands. Most people run their first query within four minutes of downloading.

How does the free tier work?

Free tier gives you 50 requests per day across all tools. Once you hit the cap, calls pause until UTC midnight. If you want unlimited, it is a one-time $9.99 payment. No subscription, no annual renewal, no surprise charges.

What happens if Reddit or LinkedIn changes their layout and it breaks?

We push an app update. When sites change, and they do, we fix it on our end and you get the update. You do not debug anything.

I use Cursor, Claude Code, Windsurf, or Zed. Does this work with my setup?

Yes. When you press Connect, the app detects which AI clients you have installed and writes the MCP config for each one. Claude Desktop, Claude Code, Cursor, Windsurf, Zed, and VS Code are all supported today.