📡 MCP Radar

zouyuanqing/dsh-vision-primitives

Native interactive visual-reasoning plugin for DeepSeek Harness: precise pixel grounding (SOM grid / zoom / annotate / measure / diff / color / OCR) + MiMo V2.5 multimodal backend, zero external MCP s

⭐ 3 stars
⑂ 0 forks
JavaScript
momentum ▲ 6.0
created 2026-08-14
on radar since 2026-08-15
ai-agentscomputer-visioncordisdeepseek-harnessdsh-pluginocrvision
View on GitHub ↗

About dsh-vision-primitives

Native interactive visual-reasoning plugin for DeepSeek Harness (DSH).

给纯文本智能体装上"精确的眼睛":以 Set-of-Mark 编号网格 + 确定性像素坐标数学 为核心,让 Harness 智能体对屏幕/图片做精确到像素的视觉交互推理 —— 全程零外部 MCP 服务器,视觉推理内核 100% 在 DSH Host 运行时内以纯 JS 执行。

Design inspired by vision-primitives-mcp, re-implemented as a native DSH plugin (dynamic Cordis plugin, Host-only package).

------ 🧠 智能体即视觉模型 插件产出图片路径 + 确定性坐标数学;Harness 多模态智能体(或内置 MiMo 后端)看图决策,插件把"模糊感知"换算成"精确像素" 🎯 SOM 编号网格 visiongrid 叠加编号网格 → visionresolve(cell) 得格子中心精确坐标,消除视觉模型坐标误差 🔍 局部无损放大 visionzoom 最近邻放大(像素级保真),保留到原帧的坐标映射链 📐 几何验证 visionannotate / visionmeasure / visiondiff / visionfindcolor / visionocr 确定性验证 🖥️ MiMo V2.5 后端 visiondescribe / visionlocate(多模态理解 + 视觉定位),并注册 mimo 模型路由(LlmAdapter,流式/函数调用/图像输入全支持) 🧱 最小 OS 边界 仅截屏 / OCR / 二进制落盘 3 类走 Host 原生 subprocess 服务调用 Windows PowerShell 系统脚本;不含桌面键鼠控制 🔒 零硬编码密钥 MiMo API key 从 DSH credentials 惰性读取,不写入源码

纯 JS 包,无 build 脚本,无需 pnpm allowBuilds 许可:

也支持 github:zouyuanqing/dsh-vision-primitives 固定版本安装。 该包同时是一个 npm 包(dsh-vision-primitives),dsh plugin add dsh-vision-primitives 也可。

visiondescribe / visionlocate 与 mimo 模型路由需要 MiMo API key:

模型:mimo-v2.5(Xiaomi 官方 API,OpenAI 兼容,1M 上下文,全模态理解)。未配置 key 时,相关工具会给出明确报错提示;纯本地视觉原语(网格/缩放/标注/测量/差分/颜色/OCR)不需要任何 key。

工具 功能 ------ visioncapture 截屏(全屏/区域,多显示器合并)或读取工作区 PNG → 会话帧 visiongrid Set-of-Mark 编号网格叠加,返回带编号图片 + 每格精确 box/center visionresolve 格子/框/点 → 帧内精确像素坐标 + 绝对屏幕坐标,记为定位锚点 visionzoom 局部无损放大(最近邻,像素级保真),保留到原帧的坐标映射链 visionannotate box/point/cross/line/text 标注绘制(视觉验证) visionmeasure 两点距离/位移/夹角,或框面积 visiondiff 帧差分(确定性变化检测):变化像素 bbox/比例 + 高亮图 visionfindcolor 颜色分割 + 连通域(CV 像素级定位,零视觉模型) vi

From the project README.

Maintaining this server?

Add the radar badge to your README — it shows your project was picked up by MCP Radar and links to this page:

[![On MCP Radar](https://mcp.liqiwa.com/badge.svg)](https://mcp.liqiwa.com/s/zouyuanqing--dsh-vision-primitives.html)

Related MCP servers

PerryLink/jevcore

TypeSafe Jev for DeepSeek Harness, the Model Context Protocol, and plain Node: typed judgments instead of prose, offline by default.

⭐ 38TypeScript
noteflowai/dsh-skills-anywhere

Reuse Agent Skills across DeepSeek Harness and MCP. Check local SKILL.md files in the browser or CI, with parser diagnostics, file hashes and clear exit codes. Try the Hugging Face playground.

⭐ 6TypeScript
pazz11/Jnpz

改配置+重启才能加 MCP?DeepSeek Harness (DSH) 插件「技能配置」解决:设置页粘贴 JSON 即连 MCP 服务器、热加载免重启;上传 zip/.skill 自动识别 SKILL.md,创建/编辑/启停技能,即点即用

⭐ 11JavaScript
hyqhyq3/dsh-mcp-manager

MCP server manager plugin for DeepSeek Harness: Settings → MCP page, OAuth (PKCE + dynamic client registration) or static-token auth, tools registered as mcp__<name>__*

⭐ 9JavaScript
Imzl-zl/dsh-mcp-manager-ui

MCP server management UI for DeepSeek Harness Web — floating panel, JSON import, and profile-backed persistence.

⭐ 3JavaScript
mrpulor-gh/dsh-nuphus-mcp

DeepSeek Harness (DSH) plugin: mount the nuphus-mcp desktop automation MCP server as native tools (mcp__nuphus-mcp__*).

⭐ 3JavaScript

📬 Get the weekly radar in your inbox

The top new MCP servers of the week, every Monday. No spam, unsubscribe anytime.