Web Eval Agent

v1.0.0Browser Automationstable

An MCP server that autonomously evaluates web applications.

debuggingdebugging-toolmcpmcp-servermodelcontextprotocol
Share:
1,240
Stars
0
Downloads
0
Weekly
0/5

What is Web Eval Agent?

Web Eval Agent is a Model Context Protocol (MCP) server that allows AI assistants like Claude, Cursor, and VS Code to mcp server that autonomously evaluates web applications.

An MCP server that autonomously evaluates web applications.

This server falls under the Browser Automation category on MCPgee, the world's largest MCP server directory with 33,000+ servers.

Features

  • An MCP server that autonomously evaluates web applications.

Use Cases

Autonomously evaluate and test web applications.
Debug web apps directly from code editors.
refreshdotdev

Maintainer

LicenseApache-2.0
Languagepython
Versionv1.0.0
UpdatedMay 20, 2026
Statushealthy
Maintenanceactive

Works with

ClaudeOpenAIwindowsmacoslinux

Installation

Manual Installation

npx web-eval-agent

Configuration

Configuration Details

Config File

claude_desktop_config.json

Performance

Response Metrics

Response Time< 200ms
ThroughputMedium

Resource Usage

Memory UsageLow
CPU UsageLow

How to Set Up and Use Web Eval Agent

Web Eval Agent MCP Server is an autonomous web application evaluation tool that uses AI-powered browser automation to test, debug, and report on web apps directly from within your code editor. Built on BrowserUse and Playwright, it navigates your application like a real user, captures screenshots at each step, records network requests and responses, and collects browser console errors — then delivers a structured report summarising what it found. It integrates with MCP-compatible clients so that AI coding assistants can validate their own generated code against a running application without manual browser testing.

Prerequisites

  • An Operative.sh account and API key — obtain free at operative.sh/mcp
  • The operative.sh CLI installed (includes the MCP server)
  • An MCP-compatible client such as Claude Desktop, Claude Code, or Cursor
  • A locally running web application or a publicly accessible URL to evaluate
1

Install the operative.sh CLI

The install script sets up the Web Eval Agent MCP server and all dependencies including BrowserUse and Playwright.

curl -LSf https://operative.sh/install.sh -o install.sh && bash install.sh && rm install.sh
2

Get your Operative API key

Visit operative.sh/mcp to create a free account and retrieve your OPERATIVE_API_KEY. This key authenticates all browser automation sessions.

3

Add the server to your MCP client config

Add the following entry to your claude_desktop_config.json (or equivalent config file for your MCP host).

{
  "mcpServers": {
    "web-eval-agent": {
      "command": "npx",
      "args": ["web-eval-agent"],
      "env": {
        "OPERATIVE_API_KEY": "your_operative_api_key"
      }
    }
  }
}
4

Set up browser authentication (optional)

If your app requires login, use the setup_browser_state tool first. It opens an interactive browser where you can sign in manually; session cookies are persisted for subsequent automated evaluations.

5

Restart your MCP client and launch your web app

Restart the MCP client. Start your application on a local port. The server exposes two tools: web_eval_agent (automated testing with URL and task inputs) and setup_browser_state (session persistence).

Web Eval Agent Examples

Client configuration

Minimal claude_desktop_config.json for the Web Eval Agent MCP server.

{
  "mcpServers": {
    "web-eval-agent": {
      "command": "npx",
      "args": ["web-eval-agent"],
      "env": {
        "OPERATIVE_API_KEY": "your_operative_api_key"
      }
    }
  }
}

Prompts to try

Sample prompts for autonomous web app evaluation using the web_eval_agent tool.

- "Test my app at http://localhost:3000 — try the full signup flow and report UX issues and console errors"
- "Navigate to http://localhost:5173 and verify that all navigation links work without 404 errors"
- "Evaluate the form validation on http://localhost:3000/contact — submit with empty fields and check error messages appear"
- "Run an end-to-end test of the shopping cart at http://localhost:3000/shop — add two items, update quantities, and report any issues"

Troubleshooting Web Eval Agent

OPERATIVE_API_KEY not found or invalid

Sign up at operative.sh/mcp to get a free API key. Paste it exactly (no extra spaces) into the env block in your MCP config. Fully restart the MCP client after saving the config file.

Agent cannot access the app at localhost

Make sure your web application is already running and serving on the specified port before prompting the agent. Test with `curl http://localhost:3000` to confirm it is reachable. The agent's browser runs on the same host machine as the MCP server.

The evaluation report is vague or misses key interactions

Write more specific task descriptions. Instead of 'test my app', use 'click the Sign Up button, fill in a test email and password, submit the form, and verify the confirmation page loads'. Concrete step-by-step task descriptions produce much more useful evaluation reports.

Frequently Asked Questions about Web Eval Agent

What is Web Eval Agent?

Web Eval Agent is a Model Context Protocol (MCP) server that mcp server that autonomously evaluates web applications. It connects AI assistants to external tools and data sources through a standardized interface.

How do I install Web Eval Agent?

Follow the installation instructions on the Web Eval Agent GitHub repository. Clone the repo, install dependencies, and add the server config to your AI client.

Which AI clients work with Web Eval Agent?

Web Eval Agent works with all major MCP-compatible AI clients including Claude Desktop, Claude Code, Cursor, VS Code (GitHub Copilot), Windsurf, and Cline.

Is Web Eval Agent free to use?

Yes, Web Eval Agent is open source and available under the Apache-2.0 license. You can use it freely in both personal and commercial projects.

Browse More Browser Automation MCP Servers

Explore all browser automation servers available in the MCPgee directory. Each server includes setup guides for Claude, Cursor, and VS Code.

Quick Config Preview

{ "mcpServers": { "web-eval-agent": { "command": "npx", "args": ["-y", "web-eval-agent"] } } }

Add this to your claude_desktop_config.json or .cursor/mcp.json

Read the full setup guide →

Ready to use Web Eval Agent?

Browse our complete directory of 33,000+ MCP servers, read setup guides for your editor, and start building with the Model Context Protocol.

33,000+ ServersFree & Open SourceStep-by-Step Guides