meshbrow-mcp

MCP (Model Context Protocol) server for Meshbrow — persistent browser infrastructure for AI agents.

Give your AI agents direct access to cloud browsers that remember logins, stay undetected, and scale to hundreds of concurrent sessions.

What It Does

The Meshbrow MCP server exposes browser automation as tools that any MCP-compatible AI agent can call:

Related MCP server: wraith-mcp

Installation

Homebrew (macOS/Linux)

brew tap meshbrow-dev/tap
brew install meshbrow-mcp

Go Install

go install github.com/meshbrow-dev/meshbrow-mcp@latest

Docker

docker pull ghcr.io/meshbrow-dev/meshbrow-mcp:latest

Configuration

Set your API key as an environment variable:

export MESHBROW_API_KEY=mb_live_your_key_here

Claude Desktop

Add to your claude_desktop_config.json:

{
  "mcpServers": {
    "meshbrow": {
      "command": "meshbrow-mcp",
      "env": {
        "MESHBROW_API_KEY": "mb_live_your_key_here"
      }
    }
  }
}

Cursor

Add to your .cursor/mcp.json:

{
  "mcpServers": {
    "meshbrow": {
      "command": "meshbrow-mcp",
      "env": {
        "MESHBROW_API_KEY": "mb_live_your_key_here"
      }
    }
  }
}

Example Usage

Once configured, your AI agent can:

"Log into HubSpot and check today's pipeline"
→ Agent calls browser_launch with a saved profile
→ Cookies are restored, already authenticated
→ Agent navigates to the dashboard and takes a screenshot

Sessions persist login state between uses — your agent logs in once and remembers it next time.

Documentation

Full guide: docs.meshbrow.dev/guides/mcp-agents

License

MIT © Bytangle Ltd