Requires browser-use  < v0.5.0 and Playwright Chromium. Currently getting an update to v0.7.6+.

​
Features

  1. Launches multiple headless (or visible) Browser-Use agents in parallel
  2. Crawls your site and records screenshots, broken links & a11y issues
  3. Works on production URLs and localhost dev servers
  4. Simple natural-language prompts via MCP in Cursor / Claude Code

​
Quick Start

# 1.  Create & activate env
uv venv --python 3.11
source .venv/bin/activate

# 2.  Install project
uv pip install -e .

# 3.  Install browser runtime once
playwright install chromium --with-deps --no-shell

​
1) Claude Code

# Register the MCP server
claude mcp add vibetest /full/path/to/vibetest-use/.venv/bin/vibetest-mcp \
  -e GOOGLE_API_KEY="your_api_key"

# Inside a Claude chat
> /mcp
# ⎿  MCP Server Status
#    • vibetest: connected

​
2) Cursor (manual MCP entry)

  1. Open Settings → MCP
  2. Click Add Server and paste:
{
  "mcpServers": {
    "vibetest": {
      "command": "/full/path/to/vibetest-use/.venv/bin/vibetest-mcp",
      "env": {
        "GOOGLE_API_KEY": "your_api_key"
      }
    }
  }
}

​
Basic Prompts

> Vibetest my website with 5 agents: browser-use.com
> Run vibetest on localhost:3000
> Run a headless vibetest on localhost:8080 with 10 agents

​
Parameters

  • URL – any https or http host or localhost:port
  • Agents – 3 by default; more agents = deeper coverage
  • Headless – say headless to hide the browser, omit to watch it live

​
Requirements

  • Python 3.11+
  • Google API key (Gemini flash used for analysis)
  • Cursor / Claude with MCP support

​
Source Code

Full implementation: https://github.com/browser-use/vibetest-use