Skills Plugins MCP Prompt Model 导航 博客 资讯 我的中心
界面增强 #deepseek-harness#dsh#dsh-plugin#plugin#sidebar#workspace

moonshile-dsh-plugins (dsh-workspace-sort)

侧边栏工作区每日按最近活动排序一次,当天顺序稳定。

moonshile @moonshile ⬇ 1 ★ 2 main

安装

dsh plugin --profile web add github:moonshile/moonshile-dsh-plugins
下载安装清单

需要可复现安装时,可在仓库后追加 #commit 固定提交。

侧边栏工作区每日按最近活动排序一次,当天顺序稳定。

该插件未提供要点说明,请参考仓库 README。

deepseek-harnessdshdsh-pluginpluginsidebarworkspace
  1. 安装并启动 DeepSeek Harness:npx @deepseek-ai/dsh web
  2. 在终端执行上面的安装命令(CLI 会解析插件并核验来源)
  3. 用 dsh plugins list 确认已安装,必要时重启 Harness 生效

插件以当前 dsh 进程的权限运行,安装时可能执行代码。请先通读仓库源码与许可证,确认无破坏性命令与越权访问;本站只做索引,不对第三方插件安全性作担保。

代码仓库github.com/moonshile/moonshile-dsh-plugins/tree/main/plugins/dsh-workspace-sort
许可证MIT
主要语言main
下载量1
GitHub 星标2
最近推送2026-09-15
收录日期2026-09-19
分类界面增强

事实信息来自公开插件目录快照(2026-10-01),介绍文案由本站再加工。

以下为插件仓库 README 全文(原始内容,由公开目录抓取整理)。

# moonshile-dsh-plugins

English | [中文](README.zh.md)

Moonshile's DeepSeek Harness (DSH) plugin collection. Each plugin lives in its own directory as an independent npm package, dependencies managed per package, orchestrated by a pnpm workspace.

[![awesome · DSH plugin](https://awesome-dsh-plugin.com/badge.svg)](https://github.com/awesome-dsh-plugin/awesome-dsh-plugin)

## Layout

```
plugins/
└── /
    ├── package.json   # standalone package: name / main / files / scripts
    ├── lib/           # implementation (host half / client half)
    ├── lib/*.test.mjs # node:test pure-function tests
    └── README.md      # English (中文见 README.zh.md)
```

## Local development

```bash
pnpm install     # install the workspace from the repo root
pnpm test        # run every plugin's tests (pnpm -r test)
```

## Publishing

Each plugin is published to npm independently (package names must be globally unique on the registry):

```bash
cd plugins/
pnpm publish --access public
```

## Plugins

Every plugin is an npm package installed with `dsh plugin` — one command, auto-activating bundle. Each plugin has its own directory and README.

| Plugin | What it does | Install | Usage |
| --- | --- | --- | --- |
| [dsh-workspace-sort](plugins/dsh-workspace-sort/README.md) | Re-sorts sidebar workspaces by last activity **once per day**; order stays stable the rest of the day | `dsh plugin --profile web add dsh-workspace-sort` | None needed — install, restart `dsh web`; it runs daily, sorting by session activity. Last-sort date is persisted at `~/.dsh/workspace-sort-state.json`. |
| [dsh-last-turn-delete](plugins/dsh-last-turn-delete/README.md) | A trash button (same icon as the pending-message queue) on the newest deletable item — the last user message or a failed `/compact` card — that **soft-deletes that turn** so it no longer enters the model context, then walks back to the previous item until a real compact boundary or none remain | `dsh plugin --profile web add dsh-last-turn-delete` | Install, restart `dsh web`, refresh. Click the trash can (confirm bubble) to delete; the button appears only after the reply finishes. Deleted turns are dimmed in place; the durable log is never rewritten. |
| [dsh-responses-replay-fix](plugins/dsh-responses-replay-fix/README.md) | Fixes OpenAI-Responses history replay rejected by strict gateways (e.g. Zhongxin `www.wxzjai.com`): assistant message ids get a `msg_` prefix and thinking items get plaintext `reasoning_text` before resend | `dsh plugin --profile web add dsh-responses-replay-fix` | None needed — install, restart `dsh web`. Targets the `zhongxin` provider by default; configure more via the settings section `responses-replay-fix.providers`. |

## Adding a plugin

1. Create `plugins//` with its own `package.json` (name, `main`, `files`, `license`, `dsh.bundle.patch`) and `lib/`.
2. Add tests as `lib/*.test.mjs` (`node:test`); verify with `pnpm test`.
3. Add a row to the [Plugins](#plugins) table.
4. Publish it to npm (`pnpm publish --access public`); the repo CI publishes it automatically on push when its version is new.

数据来源:公开的 DeepSeek Harness 插件目录与各插件 GitHub 仓库。本站为独立第三方目录,与 DeepSeek、幻方(High-Flyer)及插件作者均无隶属或背书关系。

每日精选 Skill 推荐,免费送到你邮箱

输入邮箱,每天接收一个精选 AI Agent 技能推荐。完全免费,持续更新。

提交后我们会发送一封确认邮件,点击邮件里的链接才会开始收信。

完全免费,取消任意时间。我们不会发送垃圾邮件。