Skip to main content

Visual Desktop Bridge - Give any AI full control over Windows

Project description

Visual Desktop Bridge (Claude Bridge)

Visual Desktop Bridge - Give any AI full control over Windows Currently designed for Claude Code and Claude Desktop.

[!NOTE] This project was formerly known as "Universal PC Control".

Overview

Visual Desktop Bridge is a "body" for your AI "brain". It allows Claude to:

  1. See your screen (using screenshots and grid overlays).
  2. Control your mouse/keyboard (using coordinate grids and hotkeys).
  3. Manage applications (launch, switch, close).
  4. Run locally on your PC while communicating securely with the cloud/MCP.

Key Features

  • Vision-First Control: Claude sees a grid overlay (e.g., "Click A5") to interact with any app, even those without accessibility API support (Spotify, Games, etc.).
  • Persistent Connection: Agents are remembered across sessions.
  • Fast & Lightweight: Built on fastmcp and optimized for low latency.
  • Safe: Includes a "Safety Sentinel" to block dangerous commands (like deleting files) unless explicitly approved.

Installation

Easy Install (Windows)

  1. Download the latest release.
  2. Run ClaudeBridgeAgent.exe.
  3. The agent will auto-register with your MCP configuration.

Manual Setup (Developers)

  1. Clone the repo.
  2. Install dependencies:
    pip install -r requirements.txt
    pip install -r requirements-local.txt
    
  3. Run the bridge server:
    python bridge_mcp.py
    
  4. Run the local agent:
    python local_agent.py
    

Usage with Claude

Once running, you can ask Claude to:

  • "Take a screenshot and tell me what you see."
  • "Open Notepad and type 'Hello World'."
  • "Click the Play button on Spotify."
  • "Organize the files in my Downloads folder."

Architecture

  • Bridge Server (bridge_mcp.py): Acts as the MCP entry point for Claude.
  • Local Agent (local_agent.py): Runs on the Windows PC, executing actual mouse/keyboard commands.
  • Vision Engine: Uses pyautogui and PIL to generate grid-augmented screenshots for the AI.

Safety

By default, "Safe Mode" is ON. Dangerous commands (powershell, file writes) require user confirmation via the desktop overlay.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

bridge_mcp-1.1.0.tar.gz (5.3 MB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

bridge_mcp-1.1.0-py3-none-any.whl (8.7 kB view details)

Uploaded Python 3

File details

Details for the file bridge_mcp-1.1.0.tar.gz.

File metadata

  • Download URL: bridge_mcp-1.1.0.tar.gz
  • Upload date:
  • Size: 5.3 MB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.13.5

File hashes

Hashes for bridge_mcp-1.1.0.tar.gz
Algorithm Hash digest
SHA256 f70a88890aa227134801049f467ee16ce53e07d858db6c33b745bc1a3d3865e3
MD5 ccef104b5527821db7de19f2fb750728
BLAKE2b-256 e6ddf3d063d5469485b9ca6d359da583df3efa3631503b6c5f037ca3eaa0f826

See more details on using hashes here.

File details

Details for the file bridge_mcp-1.1.0-py3-none-any.whl.

File metadata

  • Download URL: bridge_mcp-1.1.0-py3-none-any.whl
  • Upload date:
  • Size: 8.7 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.13.5

File hashes

Hashes for bridge_mcp-1.1.0-py3-none-any.whl
Algorithm Hash digest
SHA256 b2fbbeb5c10c1bb8f131e72d83e1fe40eb7f4e9981d744a58a6fadf075185e46
MD5 9f5e74acc45b2ec3e59b722bb1d7cd03
BLAKE2b-256 6c20c7181317e28afa633e649c34253cea57d9fd92e17cda2702f3bcab2a6b48

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page