Altairpaca/dsh-computer-use-windows

用于DeepSeek Harness(DSH)的Windows电脑操作:窗口绑定截图/OCR/点击及验证循环,纯OCR模式,可插拔视觉模型。

Project Overview项目介绍

This is an experimental native DSH plugin for Windows computer automation, built exclusively for DeepSeek Harness to enable more reliable desktop control. It addresses common failure modes of coordinate-only automation, where clicks can land on wrong targets due to OCR drift or overlapping unrelated windows. The plugin treats every action as an observable state transition instead of a blind coordinate command, implementing core design invariants like window-scoped coordinates, text-first positioning, and post-action verification. To get started, users can clone the repository and run the included health check script to verify their local environment meets all requirements.

The plugin implements a full set of computer-use capabilities including window-scoped screenshots with coordinate metadata, Windows OCR with word coordinates and fuzzy matching, text-guided clicking with verification and bounded retries, basic mouse and keyboard input, window enumeration and focusing, and batch action execution. It supports two operating modes: OCR-only mode that keeps all processing local without sending screenshots to any remote model, and an optional vision mode that connects to an OpenAI-compatible external VLM endpoint. All API credentials are pulled from environment variables or host credential storage, and never stored in the plugin configuration. To run a local interactive smoke check, execute ./scripts/check-health.ps1 in PowerShell on your target Windows workstation.

The project is currently in experimental alpha, suitable for development and controlled testing but not yet ready for production-grade unattended desktop automation. It requires Windows 11 (recommended), PowerShell 7.4 or newer, Node.js 20 or newer, and the appropriate Windows OCR language pack for your use case. The repository includes hosted GitHub Actions CI that runs static checks on Windows latest, including JavaScript syntax validation for the DSH plugin wrapper, PowerShell parser checks for the Windows implementation, and entry point existence validation. It is released under the open-source MIT license, and users are reminded to only use it on systems they are authorized to automate.

这是面向DeepSeek Harness(DSH)的原生实验性Windows电脑控制插件,核心设计围绕窗口作用域感知、OCR文本定位、动作后验证和失败重试展开,解决了传统坐标式控制容易因窗口错位、OCR偏移导致点击出错的脆性问题。它支持无需外部视觉大模型的纯OCR模式,也可接入兼容OpenAI接口的可选视觉模型。

本插件提供了截图、OCR识别、文本点击、键鼠输入、窗口管理等一系列电脑控制能力,所有操作都将每个动作视为可观测的状态转换而非盲目的坐标指令,遵循目标窗口绑定、文本优先定位、动作后验证等设计原则。适合DSH开发者测试Windows平台的桌面自动化场景,也适合对自动化可靠性有要求的实验性工作流。

本项目目前处于实验性Alpha阶段,推荐在Windows 11系统使用,需要PowerShell 7.4+、Node.js 20+以及对应语言的Windows OCR语言包。可运行本地健康检查脚本验证环境,代码以MIT许可证开源,仅可用于授权操作的系统,开启视觉模式时截图会发送到配置的第三方端点。

Pre-install check安装前体检Compatibility · Security兼容性 · 安全性 1 warning1 项注意
  • Only 3 stars - very few users, little community feedback星标只有 3,几乎没人在用,遇到问题缺少社区反馈
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:Altairpaca/dsh-computer-use-windows

把 Altairpaca/dsh-computer-use-windows 加入你的 DSH 配置(web profile)即可启用。

READMEREADME

DSH Computer Use for Windows

Experimental Windows computer-use bundle for DeepSeek Harness, built around window-scoped perception, text-grounded actions, and post-action verification.

The repository grew out of a real desktop-automation failure mode: coordinate-only control was brittle when screenshots included unrelated windows, OCR positions drifted, or a click silently landed on the wrong UI state. The implementation therefore treats every action as an observable state transition rather than a blind coordinate command.

中文简介:面向 DeepSeek Harness 的 Windows computer-use 实验插件。核心是目标窗口绑定、OCR 文本定位、点击后验证与失败重试;视觉模型是可选项,纯 OCR 模式不需要外部 VLM。

Status

Experimental alpha. The repository contains a real DSH plugin wrapper (plugins/index.js), helper runtime (helper/cu.ps1), bundle patch, skill documentation, local health checks, and hosted Windows static CI. It is suitable for development and controlled testing, but the project does not yet claim production-grade unattended desktop automation.

The remaining release gate is a clean-install / real-DSH validation matrix on representative interactive Windows configurations.

Design invariants

Invariant Why it exists
Window-scoped coordinates screenshots, OCR results, and clicks must refer to the same target-window coordinate system
Text before coordinates when text is observable, click_text resolves the target from OCR instead of asking the model to guess pixels
Verify after action a click is successful only when the expected post-action state can be observed
Retry with evidence offset retries return the attempted positions and verification result instead of hiding failure
Vision is optional the deterministic OCR path remains usable without sending screenshots to an external model
Credentials stay external model/API credentials are read from environment or host credential storage, not committed config

Implemented surface

Capability Current surface
computer_screenshot full-screen or target-window screenshots with coordinate metadata
computer_ocr Windows OCR with word coordinates, filtering, and fuzzy query support
computer_click_text OCR locate → click → verify → bounded offset retry
computer_mouse / computer_keyboard mouse, drag, scroll, keyboard, and clipboard-oriented input primitives
computer_window enumerate, focus, and resolve target windows
computer_use_run batch action execution through one tool call
computer_vision optional pluggable OpenAI-compatible vision endpoint
computer_calibrate DPI / residual calibration support

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-docker 下一个 Next dsh-conv-search →