DeepSeek Harness Plugin Hub

Publish and manage complete Harness Profiles. Discover Plugins for your next setup.

Explore

PluginsPresetsDocsNews

Community

Publish a pluginContactReport an issue

Resources

Plugin Hub on GitHubDeepSeek HarnessSystem statusPrivacy notice
© 2026 DeepSeek Harness Plugin HubPowered byPaxTech

Independent and unofficial. Not affiliated with, authorized by, or endorsed by DeepSeek.

Pixel Art — DSH Plugin for DeepSeek Harness
← Plugins
P

dsh-pixel-art

Pixel Art

Handwritten pixel-art skill plugin: teaches the agent to use scripts to draw any image the user requests pixel by pixel, including recreating images from references through downsampling, and requires it to revise until satisfied before delivering. Handwritten pixel-art skill for DeepSeek Harness.

The plugin will be installed here. Keep web if you are unsure.

npx -y @deepseek-ai/dsh plugin --profile web add github:Alkaid4521/dsh-pixel-art#2897b33f89e823239b0a4b07eed3a342687a75bf
READMECompatibilityVersions

Compatibility and provenance

Pixel Art is published as dsh-pixel-art and currently resolves to version 0.1.1. The Hub verifies its manifest and preserves the exact installation source for reproducible installs.

DSH compatibility
*
Runtime surfaces
any
Release source
github
Registry updated
9/30/2026

Versions

0.1.1stable
9/30/2026

Related plugins

Loading related plugins…

Latest
0.1.1
DSH
*
HMR
Process restart
Tree shaking
Safe tree shaking not declared
Unpacked size
Unavailable
Files
Unavailable
Surface
any
License
MIT
Source
github
GitHub
★ 0
Weekly downloads
0
Last push
9/30/2026
View source ↗
README badge

Click the badge to copy Markdown for your README.

Do you maintain this Plugin?Claim benefit · Priority security scan

Verify the GitHub repository declared in package.json to manage this listing. After you claim it, Hub will prioritize a security scan of the current version and publish the result when it passes.

Claim this Plugin →
Report an issue
DeepSeek Harness Plugin Hub
ProfilesPluginsCategoriesNewsDocsSign inManage Profiles
ProfilesPluginsCategoriesNewsDocsSign in

Related plugins

More verified plugins in vision-media.

Modlens@liustack/modlensPlug-in vision for text-only LLMs, powered by the free Antigravity CLITool Describe Image@linxin666/dsh-tool-describe-imageModel-facing describe_image tool for the dsh web GUI: gives a text-only model image understanding by asking a vision-language model at an OpenAI-compatible endpoint to describe one image (local path, http(s) URL, or attachment reference). Hot-pluggable — Vision Routerdsh-vision-routerEyes for text-only DeepSeek Harness agents: built-in free vision chain + pixel-level tools. Remote vision providers, including the default free fallback, receive image content unless strict local-only use is configured.Vision Toolkit@anionex/dsh-vision-toolkitDeepSeek Harness-native integration for agent-vision-toolkit: image Q&A, OCR, grounding, UI restoration, pixel diff, Artifacts, and Web UI.

README

dsh-pixel-art

一个 skill 型 DSH(DeepSeek Harness)插件:不规定"能画什么",而是教 agent 怎么手写像素画出用户要的任何图 —— 写脚本逐像素算颜色,想要什么画什么,不联网、不依赖绘图库、同参数必然复现。

技能名:handwritten-pixel-art

它解决什么

把画图能力做成"固定词表的程序化渲染器"时,用户一说"蓝色的鲸鱼"就露馅——词表里没有的东西一律画不出来。而实际能把任意需求画出来的是手写像素:写脚本逐像素算颜色。

这个技能把两次实战(按参考图复刻壁纸、从零画星空风景)的经验沉淀成可复用的流程与配方,并带上交付纪律:改到自己满意才能交付。

安装

方式一:从 GitHub 装

# 在 DSH 主程序目录(含 dsh.cmd)执行
dsh plugin --profile web add github:Alkaid4521/dsh-pixel-art

国内直连 github / codeload 常超时,可让 git 走镜像传输(写进 package.json 的仍是官方地址):

$env:GIT_CONFIG_COUNT=1
$env:GIT_CONFIG_KEY_0='url.https://ghfast.top/https://github.com/.insteadOf'
$env:GIT_CONFIG_VALUE_0='https://github.com/'
dsh plugin --profile web add github:Alkaid4521/dsh-pixel-art

备用镜像:gh-proxy.com、ghproxy.net(把上面 URL 里的域名换掉即可)。

方式二:本地源码 link

git clone https://github.com/Alkaid4521/dsh-pixel-art
cd dsh-pixel-art
dsh plugin --profile web add link:.

或把源码放到固定的插件源目录再 link(勿放 Downloads,防止被清理后链接失效):

#   %DSH_HOME%\plugin-src\dsh-pixel-art → dsh plugin --profile web add link:<该绝对路径>

方式三:离线分发包

下载 Releases 里的 zip 解压,进目录执行:

powershell -NoProfile -ExecutionPolicy Bypass -File .\install.ps1     # 只想看会做什么:加 -DryRun
powershell -NoProfile -ExecutionPolicy Bypass -File .\verify.ps1      # 出厂校验:哈希 / 语法 / 清单一致性 / 资源 / 体量

装完必须重启 DSH(dsh web),技能才会进会话目录;dsh --profile web --dump-config 能确认配置树里出现 dsh-pixel-art。

用法

装完直接说人话,不需要斜杠命令;agent 按技能描述自动加载,也可显式 skill(name="handwritten-pixel-art")。

你说会发生什么
画一幅山水画加载技能 → 写脚本逐像素作画 → 出 PNG → 自己看图、按指标挑毛病 → 改 → 重渲,直到满意才交
复刻这张图(附图)走技能 §10 复刻流程:读原图像素 → 降采样到目标网格 → 量化 → NEAREST 放大
画成像素风 / 平滑风像素风 = 网格 = 输出 ÷ 4,32 级量化 + Bayer 抖动;平滑风 = 网格 = 输出 ÷ 2,盒式降采样

画图需要什么环境

两条路任选一条,都不联网:

路线需要适用
首选Python 3 + Pillow画布 / 噪声 / 色标 / 量化抖动 / 自评指标一站齐全
兜底Node(零依赖,自写 PNG 编码器)机器上没有 Python + Pillow 时

两个已知坑:

  • PATH 上的 python 若是 WindowsApps 别名,运行会报 Python was not found。用完整解释器路径,例 %LOCALAPPDATA%\Python\bin\python.exe。
  • Pillow 要在:pip install pillow。

仓库结构

skills/handwritten-pixel-art/
├── SKILL.md                     唯一会被整篇读进上下文的文件:铁律 / 工作流 / 图层顺序 /
│                                逐层配方 / 立体感四件事 / 出图前处理 / 自评六指标 /
│                                交付闸 / 失败症状索引 / 复刻流程 / 维护纪律(含体量预算)
└── assets/
    ├── pixel_kit.py             起手工具包(Python + Pillow):画布 / 噪声 / 色标 /
    │                            自评指标 / 存 PNG,可直接改
    └── png_node.mjs             纯 Node 零依赖版:自写 PNG 编码器 + 画布 + 噪声 + 量化
tools/check-size.mjs             技能体量守门:SKILL.md 超预算即报错

分层原则:判据进 SKILL.md,代码进 assets/(按需读,不读不花 token)。技能不提供案例库 / 模板库 / 成品图——给样例会把后续发挥锚定在照抄上;可复用的只有判据与工具。

版本

  • v0.1.1(2026-09-29):修 pixel_kit.py 取整缺陷 —— Python round() 是银行家舍入,坐标落在 x.5 时会漏掉整列(圆、光晕只写到隔一列上,出成片的 2px 竖条伪影);Canvas.put/mix/add/get 改用 floor(v+0.5),并在 SKILL.md 症状索引补一条。
  • v0.1.0(2026-09-28):首版。

许可证

MIT