kunjinkao-os/dsh-mobile-gui-agent 预览 preview

kunjinkao-os/dsh-mobile-gui-agent

DeepSeek Harness 的 Android 移动端 GUI 代理插件,支持 ADB 控制、迭代验证、审批以及 Web 移动视图。

Project Overview项目介绍

This is a DSH-native plugin built specifically to enable DSH to control Android devices over Android Debug Bridge (ADB). Once installed, it adds a dedicated mobile_gui_agent entry to the DSH web user interface, and runs all tasks through a structured observe-decide-act-verify loop. Before installing, you need to connect and authorize your Android device via USB or wireless ADB, then verify the device is visible by running adb devices -l in your terminal. After installation, you can confirm the plugin is correctly loaded by running dsh --profile web --dump-config and checking for a dsh-mobile-gui-agent entry in the output.

The recommended workflow for this plugin is straightforward. Start by sending a short initialization message, like “Prepare a mobile task”, in the main DSH conversation. Next, select the mobile_gui_agent conversation tab, pick your connected Android device from the list, and enter your desired mobile command only into the Task field inside the mobile_gui_agent panel. It includes built-in protection against stale UI elements, adaptive screen stabilization, stuck detection, time limits for steps, and recovery for common ADB errors. It also includes approval checks for sensitive actions like sending payments, changing security settings, or installing apps, so you always stay in full control of what the agent does.

This plugin requires Node.js version 22.19.0 or newer, and a properly configured ADB installation on your host machine to work. It works with both physical Android devices and emulators, and supports both USB and wireless ADB connections. It is released under the open source MIT license, so you can use it for free for personal or commercial projects. It does not require an Android accessibility service, but note that custom Canvas, WebView, game, or image-only controls may not be properly exposed via UIAutomator, so you may need to pair it with a vision-capable LLM.

这是一个原生适配 DeepSeek Harness(简称 DSH)的插件,专门用于通过 Android 调试桥(ADB)控制已经连接的 Android 实体设备或者模拟器。它会在 DSH 的 Web 界面中新增一个 mobile_gui_agent 入口,通过观察-决策-执行-验证的闭环流程自动完成所有指定的移动端交互任务,遵循 DSH 官方清单规范打包为 npm 包,用户可直接通过 DSH 插件命令一键完成安装。

典型工作流程为,先在 DSH 普通对话中发送简短初始化消息,比如发送「准备一个移动端任务」,再选择 mobile_gui_agent 对话标签页,选中已连接的设备后在任务栏输入要执行的移动端操作指令,点击开始后即可按验证步骤自动执行。适合需要自动化测试 Android 应用、完成各类移动端交互任务的开发测试人员使用。

本插件依赖 Node.js 22.19.0 以上版本、本地已正确配置好 ADB 的开发环境,支持有线和无线两种 ADB 连接方式,采用 MIT 许可证免费开源。本插件不依赖 Android 无障碍服务,不过自定义画布、WebView、游戏控件等可能无法被 UIAutomator 识别,需要配合视觉模型使用。

Pre-install check安装前体检Compatibility · Security兼容性 · 安全性 1 note1 项提示
  • 11 stars - an early-stage project星标 11,属于早期项目
DSH walks through these 9 checksDSH 会逐条核对这 9 项

Compatibility兼容性

  • DSH, Node, OS and profile requirementsDSH 版本 / Node 版本 / 操作系统 / profile 是否满足要求
  • External dependencies and runtimes (Electron / Python / Docker, ...)外部依赖与运行时(Electron / Python / Docker 等)是否齐备
  • Conflicts with installed plugins: command names, skill / tool names, ports, duplicate MCP registration与已装插件是否冲突:命令名、skill / tool 重名、端口占用、重复 MCP 注册

Security安全性

  • Repo matches the facts registered here; archived or abandoned?仓库是否与页面登记一致,是否归档或长期停更
  • Safety of preinstall / install / postinstall and install.sh / setup.ps1preinstall / install / postinstall 与 install.sh、setup.ps1 是否安全
  • curl|bash, download-then-execute, obfuscation, unrelated domains → stop immediatelycurl|bash、下载即执行、混淆代码、无关域名 → 立刻停止
  • Typosquatting or unmaintained packages among the new dependencies新增依赖里有没有 typosquatting 或无人维护的包
  • Requested permissions vs. what the feature actually needs申请了哪些权限、是否超出功能所需(filesystem / network / shell / clipboard)
  • Any sudo / admin requirement, plus uninstall and rollback是否要求 sudo / 管理员权限,以及卸载与回滚方式

Anything uncertain must be marked unknown with a note on how to confirm it. This site's signal screen is a static snapshot, not a security audit.拿不准的必须标「未知」并说明要我怎么确认。本站的信号筛查是静态快照,不能替代安全审计。

Or use CLI install (for developers)或使用命令行安装(适合开发者)

CLI Install命令行安装

dsh plugin --profile web add github:kunjinkao-os/dsh-mobile-gui-agent#v0.2.1

把 kunjinkao-os/dsh-mobile-gui-agent 加入你的 DSH 配置(web profile)即可启用。

READMEREADME

DeepSeek Harness Mobile GUI Agent

Mobile GUI Agent cover

CI awesome · DSH plugin DeepSeek Harness Android License: MIT

English | 中文 | Changelog

dsh-mobile-gui-agent is an installable DeepSeek Harness plugin for controlling Android devices through ADB. It adds a mobile_gui_agent entry to the Harness Web UI and drives every task through an observe → decide → act → verify loop.

The repository is a single publishable npm package. The bundle patch inserts one Cordis plugin row; that plugin composes the ADB Provider, the Phone Agent Consumer, the Phone tools, the Typert Remote adapter, and the browser client under one lifecycle.

Quick start

Connect and authorize an Android device, then install the pinned release into the Harness Web profile:

adb devices -l
dsh plugin --profile web add github:kunjinkao-os/dsh-mobile-gui-agent#v0.2.1
dsh --profile web --dump-config
dsh --profile web

The device row must report device. The configuration dump must contain a # == dsh-mobile-gui-agent layer and one dsh-mobile-gui-agent row. For the clearest workflow, send one short ordinary message in the Harness conversation first, then select mobile_gui_agent and enter the real phone command in its Task field. Do not enter the phone command in the ordinary Harness conversation composer. See Requirements and Use before operating an account or a device containing personal data.

Recommended workflow

  1. Select a workspace and send a short initialization message in the ordinary Harness conversation, for example Prepare a mobile task.
  2. Select the mobile_gui_agent conversation tab.
  3. Select the connected device and enter the actual phone command only in the Task field inside mobile_gui_agent.
  4. Select Start and follow the verified steps. Do not send the phone command through the ordinary conversation composer.

Showing the opening section of the README — the full document lives in the repository以上为 README 开头摘要,完整文档在仓库内 · View the full README on GitHub →在 GitHub 查看完整 README →

← 上一个 Prev dsh-browser 下一个 Next dsh-grok-tui →