Spec Driven Development
MCP Server that facilitates spec-driven development workflows, not just Vibe Coding.
作者提供的说明(英文原文)
KDE Plasma 6 Wayland GUI automation for AI agents: 30 tools (EIS input, screenshots, AT-SPI2).
这三项永远不会合并成一个分数:意见、测量和人气是不同的东西。
我们还没有为它写通俗的介绍。下面是我们从注册中心读取到的事实。
选择你使用的 AI 应用,然后复制设置。
作者没有在注册中心列出安装信息。请查看下方作者的 README。
它在你的电脑上运行;安装后,工具列表会显示在你的 AI 应用中。
我们检查的是任何人都能看到的信息;我们没有阅读代码。不适用的检查项不计入总数。 · 检查于 2026年11月6日
注册名称已通过 GitHub 账户 VibeProgramm 验证,该账户拥有此仓库
我们还没有从 GitHub 读取这个仓库
我们还没有从 GitHub 读取这个仓库
它在你的电脑上运行,没有互联网端点
在你的电脑上运行,改用本地设置
它在你的电脑上运行;我们只能读取互联网端点的工具列表
与注册中心里这个分类中所有活跃的服务器比较。只有当发布者拥有该仓库时,星标数才计入。
它在 1 项适用的基础检查中通过了 1 项。我们没有阅读代码。在添加密钥或令牌之前,请先看看这个服务器要求哪些权限,并使用只具备所需权限的密钥。
注册中心没有列出必需的密钥,不过有些服务器会在首次使用时要求你登录。
它在你的电脑上运行。你的 AI 应用会在需要时启动它,数据不会经过作者的服务器(除非它自己调用在线服务)。
GitHub 仓库最后更新于–,最新版本(0.9.0)于2个月前进入注册中心。
支持 MCP 的应用,例如 Claude Desktop、Claude Code、Cursor 和 VS Code。
同一分类中有:Spec Driven Development, Remote Desktop Commander, Safe Fetch
同一分类中其他作者的服务器,按星标数排序。
如何看卡片:★ GitHub 星标i三大支柱:评价 · 检查 · 使用情况i更新 = 最新版本发布到注册中心的时间i
MCP Server that facilitates spec-driven development workflows, not just Vibe Coding.
Remote Desktop Commander: use files, terminal and processes on your own computers from any AI app.
MCP server for fetching URLs, safe against SSRF, DNS rebinding, and redirect-to-internal attacks.
按原文从作者的仓库复制。PopNx MCP 没有审查或背书其内容。HTML、图片和脚本已被移除。链接会离开本站。
Model Context Protocol server for Linux desktop GUI automation on KDE Plasma 6 Wayland
Actively maintained fork of isac322/kwin-mcp. This fork ships the fixes users were waiting for: MCP Python SDK 2.x support, text input in virtual sessions (host keymap isolation), reliable window focus, bounded startup waits, silent input-drop fix, honest error contracts — see the changelog and releases. Install:
uv tool install git+https://github.com/VibeProgramm/kwin-mcp
A Model Context Protocol (MCP) server that enables AI agents (Claude Code, Cursor, and other MCP clients) to launch, interact with, and observe any Wayland application in a fully isolated virtual KWin session -- without affecting the user's desktop. It also supports live desktop automation by connecting to an existing KWin session (real desktop or container) for collaborative workflows. With 30 MCP tools covering mouse, keyboard, touch, clipboard, accessibility tree inspection, screenshot capture, and window management, kwin-mcp provides everything needed for end-to-end GUI testing and desktop automation on Linux.
dbus-run-session + kwin_wayland --virtual sandbox. Your host desktop is never affected.systemd-nspawn) for collaborative "share my screen" workflows.libei protocol.Run end-to-end GUI tests for KDE/Qt/GTK applications in headless isolated sessions. kwin-mcp launches each app in its own virtual KWin compositor, interacts via mouse, keyboard, and touch input, then verifies results through screenshots and the accessibility tree -- all without a physical display.
Let AI agents like Claude Code autonomously operate desktop applications. The agent reads the accessibility tree to understand the UI, performs actions through 30 MCP tools, and observes the results via screenshots -- creating a complete feedback loop for any Wayland application.
Connect to your real desktop session and let Claude observe and interact with what you see. Use session_connect or pass --default-live-session to make live mode the default. Also supports attaching to KWin running inside containers (e.g. systemd-nspawn) for isolated agent desktops.
Integrate Linux desktop GUI testing into CI/CD pipelines. kwin-mcp's virtual sessions require no X11 or physical display server, making it suitable for headless environments like GitHub Actions or GitLab CI runners on Linux.
Automate kiosk interfaces and embedded Linux desktops running KDE Plasma or a bare KWin Wayland compositor. Use session_start for isolated virtual testing of kiosk UIs, or session_connect to attach directly to a live kiosk or embedded device session for real-time automation and diagnostics.
Requires KDE Plasma 6 on Wayland. See System Requirements for details.
1. Install
# Using uv (recommended)
uv tool install git+https://github.com/VibeProgramm/kwin-mcp
git clone https://github.com/VibeProgramm/kwin-mcp.git cd kwin-mcp uv tool install .
2. Configure Claude Code
Add to your project's .mcp.json:
{
"mcpServers": { "kwin-mcp": { "command": "uvx", "args": ["kwin-mcp"] } } }
3. Use it
Ask Claude Code to launch and interact with any GUI application:
Start a KWin session, launch kcalc, and press the buttons to calculate 2 + 3.
Claude Code will autonomously start an isolated session, launch the app, read the accessibility tree to find buttons, click them, and take a screenshot to verify the result.
The fastest way to wire kwin-mcp into your editor is to install one of the bundled plugins. Each plugin auto-registers the MCP server and ships the kwin-desktop-automation skill, which teaches the agent which tool to call when (session-mode selection, the observe → act → verify loop, US-QWERTY vs Unicode typing, AT-SPI2 coordinate translation, and other platform pitfalls).
Claude Code — install the plugin from the marketplace:
/plugin marketplace add VibeProgramm/kwin-mcp
/plugin install kwin-mcp@kwin-mcp
OpenCode — add the npm plugin to your opencode.json:
{
"$schema": "https://opencode.ai/config.json", "plugin": ["@isac322/kwin-mcp-opencode"] }
For the full integration guide (manual fallback, customising the skill, troubleshooting), see docs/ai-agent-integration.md.
Add to your project's .mcp.json:
{
"mcpServers": { "kwin-mcp": { "command": "uvx", "args": ["kwin-mcp"] } } }
Or if installed globally:
{
"mcpServers": { "kwin-mcp": { "command": "kwin-mcp" } } }
Add to your claude_desktop_config.json:
{
"mcpServers": { "kwin-mcp": { "command": "uvx", "args": ["kwin-mcp"] } } }
# As an installed script
kwin-mcp
python -m kwin_mcp
kwin-mcp-cli
kwin-mcp --default-live-session kwin-mcp-cli --default-live-session
| Tool | Parameters | Description |
|---|---|---|
session_start | app_command? str, screen_width? int (0 = auto-detect), screen_height? int (0 = auto-detect), enable_clipboard? bool (false), keep_screenshots? bool (false), isolate_home? bool (false), keep_home? bool (false), env? dict | Start an isolated KWin Wayland session, optionally launching an app. A screen size of 0 auto-detects the bounding box of the logical desktop (all enabled outputs / connected monitors). Set enable_clipboard=true to enable clipboard tools (requires wl-clipboard). Set keep_screenshots=true to preserve screenshot files after session_stop. Set isolate_home=true to create a temporary HOME with isolated XDG directories (config, data, cache, state), preventing apps from reading/writing host user settings. Set keep_home=true to preserve the isolated home directory after session_stop. Pass extra environment variables via env. |
session_connect | dbus_address? str, wayland_display? str, keep_screenshots? bool (false) | Connect to an existing KWin session (real desktop or container). Defaults to $DBUS_SESSION_BUS_ADDRESS and $WAYLAND_DISPLAY. Clipboard is always enabled. session_stop only disconnects without killing KWin or pre-existing apps. |
session_stop | _(none)_ | Stop the session and clean up. For virtual sessions: terminates KWin and all apps. For live sessions: disconnects without killing KWin or pre-existing apps. |
| Tool | Parameters | Description |
|---|---|---|
screenshot | include_cursor? bool (false) | Capture a screenshot of the virtual display (saved as PNG, returns file path) |
accessibility_tree | app_name? str, max_depth? int (15), role? str | Get the AT-SPI2 widget tree with roles, names, states, and screen coordinates (translated from window-local coordinates via compositor-side window geometry). Use role to filter to specific element types (e.g. "button", "check box"). Non-matching elements are hidden but their children are still traversed. |
find_ui_elements | query str, app_name? str, states? list[str] | Search for UI elements by name, role, or description (case-insensitive). Returns screen coordinates suitable for mouse_click and friends. Optionally filter by AT-SPI2 states (e.g. ["focused"], ["active", "visible"]). query can be empty when filtering by states only. |
| Tool | Parameters | Description |
|---|---|---|
mouse_click | x int, y int, button? str ("left"), double? bool, triple? bool, modifiers? list[str], hold_ms? int (0), screenshot_after_ms? list[int] | Click at coordinates. Supports left/right/middle, single/double/triple click, modifier keys (e.g. ["ctrl", "shift"]), and long-press via hold_ms. |
mouse_move | x int, y int, screenshot_after_ms? list[int] | Move the cursor (hover) to coordinates without clicking |
mouse_scroll | x int, y int, delta int, horizontal? bool, discrete? bool, steps? int (1) | Scroll at coordinates. delta positive = down/right, negative = up/left. Use discrete=true for wheel ticks, steps to split into smooth increments. |
mouse_drag | from_x int, from_y int, to_x int, to_y int, button? str ("left"), modifiers? list[str], waypoints? list[[x,y,dwell_ms]], screenshot_after_ms? list[int] | Drag from one point to another with smooth interpolation. Supports custom waypoints for complex drag paths. |
mouse_button_down | x int, y int, button? str ("left") | Press a mouse button at coordinates without releasing. Use with mouse_button_up for manual drag control. |
mouse_button_up | x int, y int, button? str ("left") | Release a previously pressed mouse button at coordinates |
| Tool | Parameters | Description |
|---|---|---|
keyboard_type | text str, screenshot_after_ms? list[int] | Type a string of text character by character (US QWERTY layout) |
keyboard_type_unicode | text str, screenshot_after_ms? list[int] | Type arbitrary Unicode text (Korean, CJK, etc.) via wtype or clipboard fallback (wl-copy + Ctrl+V). Requires wtype or wl-clipboard installed. |
keyboard_key | key str, screenshot_after_ms? list[int] | Press a key or key combination (e.g., Return, ctrl+c, alt+F4, shift+Tab) |
| keyboard_key_down | key str | Press and hold a key without releasing. Useful for hold
只显示了开头部分;其余内容请在仓库中阅读