deepseek-harness
deepseek-ai
DeepSeek Harness: Everything is a Plugin.
PROJECT TOPICS
INSTALL REFERENCE
dsh plugin --profile web add github:zepeng-jin/dsh-particle-whale
该命令指向仓库当前默认分支;尚无绑定当前 commit 的完整验证结果。
PROJECT README
A native Three.js 3D particle whale background plugin for DeepSeek Harness (DSH) and DSH Desktop.
Recreated with authentic WebGL shaders, full-screen forward fluid locomotion, volumetric 3D particle distribution, and real-time reactive state hooks.
1.05x scale, 1.1x speed) with subtle tail undulation.0.45x scale), and orbits with high-energy bioluminescent pulse waves.Cmd+N / Ctrl+N or clicking the "New Chat" button). Casual clicks are ignored to keep reading undisturbed.backdrop-filter: blur(20px)) to the chat input card so the whale swims visibly underneath without affecting text contrast.pointer-events: none overlay, precomputed jitter buffers, instanced mesh rendering, and decoupled async state polling.dsh-particle-whale/
├── package.json # Cordis client injection manifest
├── cordis.patch.yml # Cordis kernel plugin patch
├── build.mjs # esbuild bundler (inlines Three.js runtime)
├── src/
│ └── client.ts # WebGL rendering pipeline, physics & UI store
├── lib/
│ ├── index.js # Host entry (empty for client-only plugins)
│ └── client.js # Production bundle (zero runtime dependencies)
├── README.md
└── LICENSE
Open the DSH Desktop web profile node_modules directory:
cd ~/Library/Application\ Support/dsh-desktop/harness/profiles/web/node_modules
Symlink this project:
ln -s /path/to/dsh-particle-whale dsh-particle-whale
Register the plugin in ~/Library/Application Support/dsh-desktop/harness/profiles/web/package.json:
{
"dependencies": {
"dsh-particle-whale": "*"
},
"dsh": {
"profile": {
"bundles": [
"dsh-particle-whale"
]
}
}
}
Launch or reload DSH Desktop (Cmd+R).
cd ~/Library/Application\ Support/dsh-desktop/harness/profiles/web
npm install dsh-particle-whale
To modify the shaders, kinematics, or UI components:
# Install build dependencies
npm install
# Compile src/client.ts -> lib/client.js
npm run build
MIT © 2026
CLASSIFICATION EVIDENCE
系统优先读取 GitHub Topics,再与站内分类词典和词根规则比对。当前命中: 无有效分类标签。