Skills Plugins MCP Prompt Model 导航 博客 资讯 我的中心
工具与能力 #chinese#curated-list#deepseek-harness#dsh#dsh-skill#skills

awesome-dsh-skills

12 个实测工程技能(git 提交、代码审查、测试优先、插件开发、排障、CHANGELOG、PR 审查、任务拆解等):每个 SKILL.md 都通过对照官方规则的格式校验与隔离环境加载冒烟。

hackerfish @hackerfish ⬇ 2 ★ 5 main

安装

dsh plugin --profile web add github:hackerfish/awesome-dsh-skills
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。

12 个实测工程技能(git 提交、代码审查、测试优先、插件开发、排障、CHANGELOG、PR 审查、任务拆解等):每个 SKILL.md 都通过对照官方规则的格式校验与隔离环境加载冒烟。

该插件未提供要点说明,请参考仓库 README。

chinesecurated-listdeepseek-harnessdshdsh-skillskills
  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/hackerfish/awesome-dsh-skills
许可证MIT
主要语言main
下载量2
GitHub 星标5
最近推送2026-08-24
收录日期2026-09-19
分类工具与能力

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)。

# 🐋 awesome-dsh-skills — Tested Skills for DeepSeek Harness

**18 original engineering skills (SKILL.md). Every skill passes a format validator mirroring the official `@deepseek-ai/dsh-skill-filesystem` rules plus an isolated-DSH_HOME load smoke test. Copy, drop in, done.**

[中文](README.zh.md) · Sibling repos: [dsh-lab](https://github.com/hackerFish/dsh-lab) · [awesome-dsh-presets](https://github.com/hackerFish/awesome-dsh-presets) · [dsh-video-studio](https://github.com/hackerFish/dsh-video-studio) · [dsh-restart](https://github.com/hackerFish/dsh-restart)

## What DSH skills are

DSH scans skill roots (`~/.dsh/skills`, `.agents/skills`, project `.dsh/skills`) and registers `SKILL.md` files into the model-visible catalog. Each skill: `name` (kebab-case) + `description` required in frontmatter; optional `whenToUse` / `metadata` / `disable-model-invocation` / `user-invocable`.

## Install

```bash
git clone https://github.com/hackerFish/awesome-dsh-skills ~/dsh-skills
mkdir -p ~/.dsh/skills
cp -r ~/dsh-skills/skills/* ~/.dsh/skills/   # all, or copy individual skill dirs
```

## Quality gates (every skill)

1. `node tools/validate-skills.mjs` — field whitelist, kebab-case names, boolean value forms, single-level structure, per official rules
2. Load smoke test in an isolated `DSH_HOME` (skills placed, harness booted clean)
3. Content rule: verified facts only; no unverified "magic prompts"

## Skills (18)

| Skill | Purpose |
|---|---|
| dsh-git-commit | Conventional commits + pre-commit self-check |
| dsh-code-review | Six-dimension structured review |
| dsh-test-first | Red→green→refactor discipline |
| dsh-doc-sync | Keep docs/changelog/versions in sync |
| dsh-plugin-dev | DSH plugin dev with verified manifest/patch structures |
| dsh-plugin-client | DSH plugin client half: slot/standard-kit registration, dual build, `__ModuleLoader__` wrap, self-test |
| dsh-plugin-i18n | DSH plugin UI localization (zh/en): locale namespaces, slot `locale`, `t()` with instant switching |
| dsh-plugin-publish | Publish a DSH plugin to GitHub + awesome-dsh inclusion: build-artifact commits, repo/topics API, PR etiquette, encoding traps |
| dsh-dependency-audit | Dependency & install-script risk audit |
| dsh-refactor-safe | Baseline-first safe refactoring |
| dsh-debug-session | DSH boot/plugin failure triage |
| dsh-changelog | Keep a Changelog discipline |
| dsh-chinese-docs | Chinese technical writing conventions |
| dsh-pr-review | PR review checklist & structured feedback |
| dsh-task-breakdown | Task decomposition with acceptance criteria |
| dsh-skill-adapter | Adapt public portable skills to DSH discovery roots, active tools, and safety constraints |
| dsh-webapp-testing | Browser-evidence web-app testing with console, network, server, and regression-test checks |
| dsh-office-artifacts | Format-aware XLSX, DOCX, PPTX, and PDF creation or repair with reopen/render verification |

See [CONTRIBUTING.md](CONTRIBUTING.md) to submit a skill. Contributors: [CONTRIBUTORS.md](CONTRIBUTORS.md)

## License

[MIT](LICENSE)

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。