WayneJin0918/dsh-wm

DeepSeek Harness 的可玩世界模型工具包:查看帧、命名 3D / 像素 / 潜在路径、测量运行,并对研究循环进行 RSI 分析。

Project Overview项目介绍

dsh-wm is a native toolkit for world model research built exclusively for DeepSeek Harness. It provides a full set of tools for inspecting world model output frames, comparing predictions to ground truth, scoring runs, and iterating on the research loop. You can install it with a single DSH plugin command: dsh plugin --profile wm add github:WayneJin0918/dsh-wm, and it comes with a built-in test fixture called sunset that works right out of the box without requiring a GPU. After installation, you can launch the DSH profile with dsh --profile wm to start using the toolkit immediately.

This toolkit is designed for world model researchers and experienced DeepSeek Harness users. A typical workflow starts with opening a built-in knowledge card to name the research route (3D display, pixel video generation, or latent prediction). Next, you run tools like wm_inspect and wm_rollout_diff to check the output frames, measure differences between predictions and ground truth, and generate an interactive HTML comparison page you can scrub through. Finally, you can use the built-in RSI loop to iterate on skills, configuration, and evaluation notes based on your results.

To run dsh-wm, you need Node.js 18 or newer, and DeepSeek Harness version 0.1.0-rc6 or a compatible release. ffmpeg is an optional dependency if you need to process video or JPEG inputs, but PNG and PPM frame directories work fine offline without it. The project is released under the open-source MIT license, so you can use it for free for any purpose allowed by the license. For your first run, you can execute npm test to try out the built-in sunset fixture without any extra configuration or GPU hardware.

dsh-wm是专为DeepSeek Harness开发的原生世界模型研究工具包,提供世界模型输出帧的检测、对比、评分和研究闭环迭代功能。用户可以通过一条命令完成安装,自带内置演示样例sunset,不需要GPU就能直接运行体验。它遵循DSH插件规范,通过DSH的插件命令添加到指定配置profile中,启动对应profile即可加载使用。

适合世界模型研究人员和DSH深度用户使用,典型工作流程是先通过知识卡片确定世界模型的研究路线,再对输出滚动结果进行检测、差异对比和可视化查看,最后通过RSI研究循环迭代优化技能和配置。内置的sunset样例可以直接用来测试整个流程,用户也可以导入自己的世界模型输出结果进行分析。

本项目依赖Node.js 18以上版本和DeepSeek Harness 0.1.0-rc6或兼容版本,可选依赖ffmpeg处理视频或JPEG输入,PNG/PPM帧目录无需额外依赖即可离线使用。项目采用MIT开源协议,目前完全免费使用,首次运行可以直接执行npm test体验内置sunset样例,不用额外配置模型或GPU资源。

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 wm add github:WayneJin0918/dsh-wm

把 WayneJin0918/dsh-wm 加入你的 DSH 配置(web profile)即可启用。

READMEREADME

DSH-WM

MIT DSH Node

A playable world-model toolkit for DeepSeek Harness — look at a strip, name the route, score the run, and iterate the research loop.

Point the agent at a rollout (or just fixtures/sunset) and ask: did the second half melt, is Sora even a world simulator, and which memory recipe is allowed to win.

🚀 One command to install | Play sunset with no GPU | Built-in WM map | RSI on skills and evals

🌐 English | 中文

World-model work inside DeepSeek Harness is more fun when the agent can see the strip, name the lineage, and measure the claim. DSH-WM is the profile bundle for that: contact-sheet inspect, a compare page (side-by-side / swipe / diff heat + action HUD), three-route knowledge (3D display / pixel video-gen / latent prediction), run scoring, and an RSI loop on skills and wm.yaml.

dsh plugin --profile wm add github:WayneJin0918/dsh-wm
dsh --profile wm

Then try: Triage fixtures/sunset. Look at first, mid, last. Is this late-horizon?

DeepSeek’s product mainline can skip world models. Harness is still the research OS — this plugin is the WM lab on top of it.

Runtime: deepseek-ai/deepseek-harness

Table of contents

Play it in 30 seconds

fixtures/sunset is an 8-frame toy strip. Early pred frames stay warm and close to GT; the second half is wiped to cool blue so late-horizon collapse is obvious. No checkpoint, cluster, or GPU.

node cli.js inspect fixtures/sunset --indices first,mid,last
node cli.js view fixtures/sunset
node cli.js diff --pred fixtures/sunset/pred --gt fixtures/sunset/gt
node cli.js diagnose "is Sora a world simulator"
node cli.js knowledge --id wm-routes

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-custom-provider-reasoning 下一个 Next dsh-workspace-tools →