MCP servers that ship as code — a repository or a package you install and run locally (usually over stdio). They have no hosted endpoint, so we cannot probe or grade them: instead each one shows its repository, install command, license, stars and when it was last updated.
Collected from GitHub repositories and npm, PyPI, Cargo, Go, Maven, NuGet and RubyGems packages tagged as MCP servers, the official MCP Registry, Docker's MCP catalog, ModelScope, Baidu MCP World, Cline and other public directories; one entry per repository. A server that also offers a hosted endpoint is listed under Remote instead. These entries are not counted in our reachability, verification or readiness statistics.
-
Windows desktop automation MCP server for Google Antigravity. Screen capture, mouse, keyboard and UI Automation with an animated activity indicator.
★ 2 -
Windows computer-use MCP server: a 1:1 replica of Anthropic's computer-use tool surface, for any MCP client.
pypi MIT -
Claude Code computer use for Windows and macOS through the accessibility tree, not screenshots.
MIT ★ 1 -
Official code for "Hybrid Computer Use: A Unified Programmatic, Semantic, and Visual Interface for General Agents"
MIT ★ 55 -
Computer use for Claude Code on Windows and macOS through the accessibility tree.
npm MIT -
MCP (Model Context Protocol) server in Go for computer automation. Uses robotgo library for desktop automation.
★ 4 -
MCP server for computer automation (mouse, keyboard, screen, windows, processes)
npm MIT -
computer-usegithub.com/MrSongP/computer-useupdated this year
Local Windows computer-use plugin for Codex and Claude Code, exposing UI Automation, screenshots, input control, app launch, and MCP tooling.
★ 2 -
Turnkey macOS Computer Use Skills + Open-Source Accessibility MCP Server + Optional ZCode Native Bridge for AI Agents
Apache-2.0 ★ 1 -
Desktop Computer Use (DCU): background Windows computer-use MCP server that controls apps without moving the real cursor or stealing focus.
MIT ★ 7 -
Open-source Computer Use MCP server for any coding agent — macOS, Windows, Linux. Accessibility-first, background input, agent pointer, zoom, your signed-in Chrome.
npm Apache-2.0 ★ 40 -
💻 Give AI models complete control of your computer (probably a bad idea)
npm MIT ★ 401 -
OpenCode Computer Use: local Codex-style visual desktop control for Windows (MCP stdio, 35 tools, DPI-correct, verified E2E)
MIT ★ 3 -
Claude-style computer use for any vision model in Cursor on macOS (MCP server: screenshots + accessibility tree + OCR)
★ 1 -
Hyprland/Wayland computer-use MCP server (screenshot/click/type + tmux/CDP/task-runner copilot channels)
MIT -
Persistent JavaScript sessions for macOS computer use. Two MCP tools, native Accessibility, screenshots, OCR, and background-first interaction.
MIT ★ 1 -
wayland-computer-use-mcpgithub.com/ziggy42/wayland-computer-use-mcpupdated this year
An MCP server for controlling Linux desktops running Wayland.
Apache-2.0 ★ 1 -
Let AI agents use your Mac's apps in the background: accessibility-tree computer use as an MCP server. Native Swift, no dependencies.
MIT -
Computer Use MCP server for any coding agent — accessibility-first desktop control on macOS, Windows and Linux, background input, agent pointer, zoom, your signed-in Chrome. Downloads the signed native binary from GitHub Releases on first run.
npm Apache-2.0 -
Windows computer-use MCP server for OpenCode, Claude Desktop and Codex CLI. Indexed UI-Automation controls, set-based tree filtering, verified clicks - no vision model, no pixel guessing.
Apache-2.0 -
linux-computer-usegithub.com/tak-uukti/linux-computer-useupdated this year
Linux/X11 computer-use tools for AI agents — Pi, Claude Code, OpenCode, any MCP-aware client. AT-SPI + xdotool, ~1k LOC. Drop-in replacement for the macOS pi-computer-use.
MIT ★ 4 -
AX-first macOS computer use for DeepSeek Harness: accessibility snapshots with stable refs, background input, Unicode typing, menus, apps, set-of-mark screenshots, on-device OCR, verified actions, and optional Jev (TypeSafe System One) semantic guards — 1
npm MIT -
Local MCP tools for AI agents to debug Firefox and control Windows applications. Codex desktop plugin and VS Code support.
MIT -
Accessibility-first computer-use framework for macOS, Windows, Linux, and the browser (CDP): pruned accessibility-tree snapshots, element refs, and a safety layer, exposed as an MCP server + CLI.
pypi Apache Software License -
computer use for ai agents on omarchy and hyprland: screenshots, mouse, keyboard, ocr, accessibility tree and window control. a claude code plugin and mcp server.
MIT -
⚡ Ultra-fast hybrid Browser & Desktop Computer-Use MCP Server (DOM + Native OS Accessibility + LiquidAI D1 Vision) with human-like physical mouse movement and Paint/canvas drawing.
MIT ★ 1 -
Computer Use (Rust): a fast, safe MCP server that lets any AI client see and operate your Windows desktop. Native Rust, UI Automation + OCR, scripted batches, MCP 2026-07-28.
Apache-2.0 ★ 3 -
Local desktop automation for MCP-capable agents on macOS, Linux and Windows. Screenshots, accessibility trees and input stay on your machine.
MIT -
Standalone MCP server that lets AI clients see and operate Windows apps: screenshots, UI Automation, keyboard and mouse. Local stdio or OAuth-protected remote HTTP.
Apache-2.0 -
a11y-computer-use: accessibility-first computer use for AI agents. Pruned a11y-tree snapshots, element refs instead of pixel coordinates, a safety layer, an MCP server and CLI. macOS, Windows, Linux, and Chromium over CDP.
Apache-2.0 ★ 1 -
macOS Computer Use MCP Server - 33 tools for screen control via Model Context Protocol. Works with Claude Code, Cursor, LM Studio, Ollama, llama.cpp, MLX, and all MCP clients.
MIT ★ 1 -
🖱️ Let Factory Droid control your Mac with the Codex app's Computer Use tool. One-command install, no approval forms, works with any Codex install path.
MIT ★ 3 -
Computer use for Grok Build: a plugin and MCP server that let the agent see the screen, click and type in desktop apps and browsers on macOS, Windows and Linux. Accessibility-tree first, with safety hooks. Built for UI testing in the dev loop.
Apache-2.0 -
MCP server letting AI agents drive a Linux/X11 desktop — AT-SPI accessibility-tree perception with element-level actions, instead of screenshot + coordinate clicking.
MIT -
macos-computer-use-skillgithub.com/wimi321/macos-computer-use-skillupdated this year
Standalone MCP server that gives AI agents full GUI control over macOS — screenshots, mouse, keyboard, apps, clipboard, and multi-display. Zero private dependencies.
MIT ★ 20 -
Drive macOS apps from Claude Code through the Computer Use engine of OpenAI's ChatGPT/Codex app — no Codex quota, zero model turns per click: script (cua_repl-style JS), batch with text targeting, diff output, macros. Includes an architecture analysis of Codex Computer Use.
MIT ★ 2 -
OpenAI Codex Computer Use for omp (oh-my-pi) and MCP clients — control macOS apps and Google Chrome with the same cua_repl runtime, Chrome tab groups and approvals that Codex and T3 Code use.
MIT -
Build like you have a team. Open-source Mac control for your AI agents: background Chrome tabs, uploads, animated pointers and a live cloud tray. MCP, CLI & JavaScript. MIT.
MIT -
Self-hosted Computer Use MCP server for Windows: 21 tools, built-in OCR (Arabic+English), multi-monitor, hard consent gate on every mutating action. Works with Claude Code/Desktop, Codex, Cursor on any plan.
MIT -
MCP server that lets Claude Code, opencode and other agents use your Mac apps in the background, without moving your mouse or taking focus. Built on agent-desktop.
Apache-2.0 ★ 2 -
linux-computer-usegithub.com/Roadmvn/linux-computer-useupdated this year
Computer use for Linux. An MCP server that lets Claude Code, OpenAI Codex or any MCP client drive a real browser on your own machine, with a live view, human takeover and safety guardrails.
MIT ★ 1 -
Computer use MCP server for Claude Code, Codex and other AI agents. Reads the UI tree instead of taking a screenshot every step, with Jev (TypeSafe System One) as a fast judge that can also look at the screen to find icons and game objects. Local macOS, local Windows, or remote Windows over SSH.
MIT ★ 4 -
Windows computer-use MCP server (26 tools) with a human-in-the-loop approval gate: physical-input-only dialogs, signed guard switches, deny/launch lists, hash-chained audit log. 70+ unit tests, green CI.
MIT ★ 5 -
windows-llm-computer-use-bridge-publicgithub.com/dermanov-ru/windows-llm-computer-use-bridge-publicactive
Let an LLM agent drive a real Windows or Linux desktop over HTTP/MCP: UI Automation & AT-SPI selectors, server-side waits, batched actions, screenshots. Computer use for Claude, Codex and any MCP agent.
MIT -
AI computer use for the whole Mac, with any AI: Claude, ChatGPT, Gemini, OpenRouter or a local model. A native MCP server, a menu bar agent that learns shareable skills, and benchmarks you can run on your own model. No API key needed through MCP.
★ 4 -
Windows Computer Use for AI Agents. Both a tool (22 MCP tools for click, type, screenshot, OCR, UI inspection) and an agent (autonomous mission engine, macro recorder, intent-based discovery, event watchers). Built with opencode (DeepSeek V4). Ships as MCP server, web UI, and Tauri desktop app.
MIT ★ 53 -
Windows computer-use MCP server - 1:1 replica of Claude Desktop's tool surface, for the Claude Code CLI or any MCP client. Screenshots, mouse, keyboard, multi-monitor. Full tier everywhere: type into IDEs, launch loose dev builds, let the agent test the app you're building.
MIT ★ 8 -
background-computer-usegithub.com/Panchangam18/background-computer-useupdated this year
MCP server + skill library for driving macOS apps in the background — click, type, scroll, and inspect accessibility trees without activating apps or moving your cursor. Works with Claude Code, Claude Desktop, Cursor, Codex CLI, Goose, and any other MCP-capable LLM client.
MIT ★ 5 -
The most precise Windows computer use for DeepSeek Harness: 19 tools address controls by exact UI Automation identity, and numbered marks are checked against the live window before input, so a stale target is refused instead of mis-clicked. Batches, incremental observations and compact receipts cut round trips and token cost.
MIT ★ 6 -
AX-first computer use for AI agents on macOS with optional Jev (TypeSafe System One) semantic guards: calibrated target/input judgments before an irreversible action, decisions kept in code. Accessibility-tree targeting, window-scoped input, clipboard-safe paste, read-back verification. Ships a pip CLI, a pi package and a DeepSeek Harness plugin.
MIT ★ 8 -
Computer-use MCP server for Linux Mint X11 desktops. wip.
★ 1 -
open-compute-mcpgithub.com/ellmos-ai/open-compute-mcpupdated this year
npm launcher for the open-compute MCP server: model-agnostic computer-use tools (screenshot capture, canonical safety-gated actions, Windows UIA element targeting) over MCP stdio. Spawns the Python open-compute server (from GitHub).
npm MIT -
Fast computer use for AI agents: event-driven settling MCP server - replaces time.sleep between clicks, waits for the UI to react and stop changing, returns the settled screenshot. No blind waits, no wasted vision turns.
NOASSERTION -
npm launcher for the open-compute MCP server: model-agnostic computer-use (screenshot, safety-gated actions, Windows UIA) over MCP; spawns the Python open-compute server from GitHub.
MIT ★ 1 -
Local MCP server for computer use that leaves your desktop alone: accessibility-first control, live browser handoff, an isolated Linux display, and human takeover.
MIT ★ 2 -
Let AI models use your Windows computer through MCP: screenshots, mouse, keyboard, windows, UI Automation and OCR, with every monitor and every window's position made explicit to the model.
NOASSERTION -
Open-source computer-use harness for any OpenAI-compatible model. Waits for the screen instead of sleeping, so agents finish faster on the same model. Includes an MCP server and RL rollouts.
Apache-2.0 ★ 1 -
Computer-use MCP server: give Claude Code, Codex, Cursor or any MCP client hands on a real screen: your computer (Mac/Windows/Linux) and your Android phone (beta). Clicks, types and reads the screen, even in apps with no API.
npm MIT -
MCP server for computer use in MacOS
MIT ★ 242 -
WinCU / Windows Computer Use / Computer Action / Windows MCP: an MCP server that lets AI models operate a Windows PC with screenshots, mouse, keyboard, clipboard, window control and UI Automation. Works with Claude, ChatGPT and other vision models.
MIT ★ 3