dsh-mall
DSH Mall — GitHub #dsh-plugin topic browser with one-click install/uninstall and publish-your-plugin. · DSH 商店:GitHub #dsh-plugin 全量目录,浏览/搜索/排序/筛选 + 一键安装卸载 + 上传入口。
安装
npx -p @deepseek-ai/dsh dsh plugin --profile web add github:hoyyang/dsh-mall说明文档
阅读完整 README ↗DSH Mall
The complete plugin mall for DeepSeek Harness — browse the entire #dsh-plugin ecosystem (11,000+ repos), one-click install, AI-reviewed smart install, five-dimension practical scores, and a full UI in 9 languages.

✨ Features
- 🌐 Complete catalog — 11,000+ GitHub repos from our own CI index (incremental every 2h + daily full rebuild), served over CDN with a bundled snapshot fallback. Never rate-limited.
- 🧠 Smart install / update / uninstall — a pre-install AI security review (install / caution / refuse) runs through your configured model; local risk lists guard uninstalls. Falls back to the regular path when AI is unavailable.
- 📊 Five-dimension practical score — maintain / practical / popularity / ease / signal, weighted geometric mean × confidence, with a radar chart and "why recommended" reasons.
- 🏷️ LLM-tagged labels & descriptions — 11,032 plugins carry Chinese function tags and one-line descriptions in 9 languages, refreshed through the index pipeline (no per-user LLM cost).
- 🌍 9-language UI — English, 中文, 日本語, 한국어, Español, Français, Deutsch, Português, Русский — one click switches the entire store.
- 🛡️ Trust badges — machine scan (dsh.bundle verified in the repo tree), awesome curated, human verified, has-skill, stale & npm-unlinked warnings.
- ⭐ Editor picks & For You — weekly curated picks; profile-based recommendations with MMR diversity plus a cold-start quiz.
- ⬇️ Signals you care about — today's +stars, npm downloads (30d & total), publish date, version capsules.
- 🧰 Full lifecycle management — one-click update, daily auto-update (03:30), rollback, enable/disable switch (hot), task panel with progress & cancel.
- 📖 Safe README rendering — sanitized markdown with shields/badge cleanup, plus parsed install commands from the README (display-only, copyable).
- 🔌 Agent-friendly — ships the
find_dsh_store_plugintool and a skill so agents can discover plugins in conversations.
📦 Install
dsh plugin add dsh-mall
# or, for a specific profile:
dsh plugin --profile web add dsh-mall
Restart dsh web and open DSH Mall in the sidebar (above Settings). You can also reach the store settings from the official Settings window.
✅ Requirements
- DeepSeek Harness (dsh web) 0.1.0-rc.8 or newer — tested against 0.1.0-rc.8.
- npm peers:
@deepseek-ai/cordis ^4.0.1,@deepseek-ai/dsh-settings ^0.1.0-rc.6(optional),@deepseek-ai/dsh-tools ^0.1.0-rc.6— these are resolved automatically when installed through the store/plugin tooling. - Optional: a GitHub token (
DSHM_GITHUB_TOKENorgithubTokenin your cordis config) raises the API quota for search/verify/version lookups. It is kept in memory only and never written to disk or sent anywhere else.
🗂 How data works
The catalog is built by our companion CI repo hoyyang/dsh-market-index: GitHub search + the curated awesome-dsh-plugin directory (1,900+ hand-picked entries), classified into 20 categories, enriched with bundle scans, npm linkage, dormant detection, README signals and LLM tags. dsh-mall reads it over raw/jsDelivr CDN channels, refreshes every 30 minutes, and falls back to a bundled snapshot when GitHub is unreachable or rate-limited.
🔐 Security
- All mutating endpoints (install/uninstall/update/toggle) accept same-origin requests only.
- Install sources are always shown before confirming; smart install routes repository content into a headless AI review as untrusted data (delimited, non-executable).
- GitHub tokens are never persisted; user-level data-source/token routes were removed (deployment config only).
- Third-party plugins are third-party code — review before installing. The machine-scan badge means installable structure, not a security audit.
🤔 Why another marketplace?
40+ marketplace plugins exist — most are similar. DSH Mall bets on audit + curation + trends + evidence: machine scan verdicts, curated picks, five-dimension scoring with explanations, LLM tags, and an independent data pipeline. If you want numbers first, this is the store for you.
📄 License
MIT © hoyyang. Data from GitHub and awesome-dsh-plugin (MIT) under their own terms.
Built with ideas from dshmarket and dsh-market.