desktop

$npx mdskill add uvwt/agentdock/desktop

Automate macOS desktop tasks and observe system state

  • Automates macOS desktop operations such as window focus, app switching, and screenshot capture.
  • Deploys macOS native APIs for automation and observation of system state.
  • Decides actions based on user requests and system state information provided by AgentDock.
  • Delivers results through notifications or direct access to the automated tasks.

SKILL.md

.github/skills/desktopView on GitHub ↗
---
name: desktop
description: macOS desktop automation as an AgentDock Skill workflow.
version: 1.0.12
---

# macOS Desktop Skill

AgentDock 的 macOS 桌面自动化能力已经从 core 工具拆分为此 Skill 工作流。

## 调用入口

- `skill_action=observe`：`action=preflight | list_apps | app_state | window_list | snapshot | snapshot_app`。
- `skill_action=act`：`action=focus | move | click | double_click | scroll | drag | type | set_value | secondary_action | hotkey | wait`。
- `skill_action=clipboard-read` / `skill_action=clipboard-write`：读写剪贴板。

## 数据目录

截图原始文件写入 `AGENTDOCK_HOME/skill-data/desktop/artifacts`；`snapshot` / `snapshot_app` 同时发布到 `AGENTDOCK_HOME/public-artifacts`，并返回包含 `artifact_id` 的轻量 `screenshot` 引用。配置 `AGENTDOCK_SERVER_URL` 时额外返回带过期时间的签名 `screenshot.url`。需要模型查看图片时，再调用 AgentDock `view_image(artifact_id=...)`；Desktop Skill 不直接返回 Base64。历史截图已统一迁入同一私有数据目录。

- 动作参数不支持别名：`skill_action=act` 必须使用 `action` 枚举值；`secondary_action` 的辅助动作名必须放在 `ax_action`；`wait` 使用 `ms`。

## 辅助脚本执行

Skill 本体是本说明文档。确需调用包内辅助脚本时，在 Skill 包根目录使用相对路径执行；运行宿主负责切换到包根目录并把所需变量注入当前子进程。

```bash
printf '%s' '{"skill_action":"<动作>"}' | python3 run.py
```

输入必须是 JSON 对象。写操作仍按本文档中的确认规则执行。

| 动作 | 用途 |
|---|---|
| `status` | Run a desktop preflight check and report desktop automation readiness. |
| `observe` | Unified macOS desktop observation action: preflight, list_apps, app_state, window_list, snapshot, or snapshot_app. |
| `act` | Unified macOS desktop action: focus, move, click, double_click, scroll, drag, type, set_value, secondary_action, hotkey, or wait. |
| `clipboard-read` | Read macOS clipboard text. |
| `clipboard-write` | Write macOS clipboard text and optionally verify by reading it back. |

More from uvwt/agentdock

SkillDescription
baidu-netdiskUse this skill when managing Baidu Netdisk files through the bdpan CLI inside the application data scope.
barkSend Bark-compatible notifications with Python standard library while keeping the device key in the private Skill data environment.
cloudsaverUse this skill when calling the local CloudSaver API for status checks, resource search, Douban hot lists, share parsing, or transfer actions.
code-debriefExplain and teach the core implementation produced by vibecoding from real code, focusing on the main call chain, core code, state and data flow, design reasoning, and transferable engineering knowledge. Use when the user asks to explain, review, learn, debrief, or understand recent code changes, a commit, PR, diff, feature, module, or code path.
codex-usageUse this skill when checking the current macOS user Codex CLI account state or Codex usage quota.
douban-marks只读查询豆瓣电影、图书、音乐标记列表和电影条目联想。用于读取用户自己的已看、想看、在看等标记；公开主页可直接读，私密列表需要本机环境或 .env 中的 DOUBAN_COOKIE。禁止用于批量采集、绕登录、写入标记或读取非授权账号。
douyin-hot获取抖音实时热榜并返回标题、热度和跳转链接。
find-skills-skillSearch and discover OpenClaw skills from various sources. Use when: user wants to find available skills, search for specific functionality, or discover new skills to install.
grok-quotaQuery detailed Grok weekly usage, GrokChat/GrokBuild breakdown, plan, monthly credits, and pay-as-you-go state from local xAI OAuth credentials.
linkwardenManage a Linkwarden instance through its official HTTP API: check configuration, search and inspect bookmarks, collections, tags and highlights, and perform controlled link, collection or tag changes with explicit confirmation for destructive actions.