mcpField
U
UI Explorer MCP Server

UI Explorer MCP Server

Local
github.com/modularflowOther0Updated 74w ago

About UI Explorer MCP Server

UI Explorer MCP Server provides tools for inspecting and interacting with on-screen UI elements, aimed at UI exploration and automation. It also supports recording and replaying workflows, either independently or through MCP.

Key features

  1. 01Inspect UI element hierarchies
  2. 02Capture screenshots with highlighted elements
  3. 03Click UI elements by coordinates
  4. 04Focus inspection on screen regions
  5. 05Record and replay UI workflows
  6. 06Record keyboard input and screenshots

Use cases

  1. 01Inspect controls in a foreground window
  2. 02Automate a UI workflow with recorded actions
  3. 03Verify UI elements using highlighted screenshots

MCP support

Exposes UI exploration and interaction tools to MCP clients, including hierarchy inspection, screenshots, coordinate clicks, and keyboard input. Macro workflows can also be recorded and played back through MCP.

Configuration

Local server config is the same for every client. Copy it into your client’s MCP config file.

No install method listed yet

We haven’t collected a package or remote endpoint for this server. Follow the instructions in the project README.

Repository

Tools

No tool list yet

Local servers expose tools only once running on your machine, and the maintainer has not provided a tool list.

Details

From the project README

UI Explorer MCP Server

An MCP server that provides tools for exploring and interacting with UI elements on your screen.

Features

  • Explore UI hierarchies: Scan and analyze all UI elements on your screen
  • Screenshot UI with highlights: Visualize UI elements with boundaries and hierarchy
  • Control mouse clicks: Click on UI elements based on coordinates
  • Explore specific regions: Focus on parts of the screen like top-left, center, etc.
  • 🎬 Macro Recording: Record and playback UI workflows independently or through MCP

🚀 Quick Start: Standalone Macro Recording

New! You can now record UI workflows without running the MCP server:

# Record a workflow
python record_macro.py --name "Login Workflow"

# Play it back
python play_macro.py --file "macros/Login Workflow.json"

Windows users can use the batch files:

record_macro.bat "My Workflow"
play_macro.bat "macros/My Workflow.json"

👉 See QUICK_START.md for a 2-minute tutorial 👉 See MACRO_TOOLS.md for complete documentation

What Gets Recorded?
  • ✅ Mouse clicks with exact coordinates
  • ✅ Keyboard input with smart text buffering
  • ✅ UI context for reliable playback
  • ✅ Screenshots for verification
  • ✅ Timing information for natural replay
Recording Controls
  • F9: Start/Stop recording
  • F10: Pause/Resume
  • ESC: Emergency stop

Installation Options

Option 1: Using pip (recommended)

Install the package globally or in a virtual environment:

pip install mcp-ui-explorer
Option 2: Using git clone
  1. Clone the repository:
git clone https://github.com/modularflow/mcp-ui-explorer
cd mcp-ui-explorer
  1. Install the package:

    • Using pip in development mode:
    pip install -e .
    
    • OR using uv (recommended for development):
    uv venv
    source .venv/bin/activate  # On Windows: .venv\Scripts\activate
    uv pip install -e .
    

MCP Server Configuration

After installing the package, you need to configure your MCP client to use the UI Explorer server.

For Roo in Cursor

Add this JSON to your MCP Server config at: C:\Users\<user_name>\AppData\Roaming\Cursor\User\globalStorage\rooveterinaryinc.roo-cline\settings\mcp_settings.json

{
  "mcpServers": {
    "ui-explorer": {
      "command": "uvx",
      "args": [
        "mcp-ui-explorer"
      ]
    }
  }
}
For Claude Desktop App

Add the server configuration to: C:\Users\<user_name>\AppData\Roaming\Claude\claude_desktop_config.json

{
  "mcpServers": {
    "ui-explorer": {
      "command": "uvx",
      "args": [
        "mcp-ui-explorer"
      ]
    }
  }
}
For Direct Use with uvx

If you've installed the package (either globally or in a virtual environment), you can directly use uvx to run the server without additional configuration:

uvx mcp-ui-explorer

MCP Tools

The UI Explorer provides several MCP tools for UI automation and exploration. These tools are available when using the MCP server interface.

Usage Guide

1. Explore UI Structure

Use the explore_ui tool to get a complete hierarchy of UI elements:

Parameters:

  • region: Screen region to analyze ("screen", "top", "bottom", "left", "right", "center", etc.)
  • depth: Maximum hierarchy depth to analyze (default: 5)
  • min_size: Minimum element size to include (default: 5px)
  • focus_window: Only analyze the foreground window (default: False)
  • visible_only: Only include elements visible on screen (default: True)
  • control_type: Type of control to search for (default: "Button")
  • text: Filter elements by text content (optional)
2. Take a Screenshot with UI Elements Highlighted

Use the screenshot_ui tool to visualize the UI elements:

Parameters:

  • region: Screen region to analyze
  • highlight_levels: Use different colors for hierarchy levels (default: True)
  • output_prefix: Prefix for output files (default: "ui_hierarchy")
3. Click at Screen Coordinates

Use the click_ui_element tool to click at specific coordinates:

Parameters:

  • x: X coordinate to click (required)
  • y: Y coordinate to click (required)
  • wait_time: Seconds to wait before clicking (default: 2.0)
4. Keyboard Input and Shortcuts

The tool also provides options for keyboard input and shortcuts. See the UI Explorer guide in the MCP interface for details.

Example Workflow

  1. First, explore the UI to understand what's on the screen:

    explore_ui(region="screen", control_type="Button")
    
  2. Take a screenshot to visualize the elements:

    screenshot_ui(region="screen")
    
  3. Note the coordinates of elements you want to click, then click at those coordinates:

    click_ui_element(x=500, y=300)
    
  4. Type text or use keyboard shortcuts as needed:

    keyboard_input(text="Hello world", press_enter=true)
    

Requirements

  • Windows operating system
  • Python 3.10+
  • MCP 1.6.0+
  • PyAutoGUI
  • PyWinAuto
  • Pillow
  • Pydantic 2.0+

Comments

Markdown supported

New to MCP?

From what the protocol is to installing your first server, these answers cover most first questions.

An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.