dshfind
The learning & sharing community for DeepSeek Harness (DSH)
Learn the principles · Discover plugins · Share best practices
🌐 dshfind.com
English | 简体中文 | 日本語 | 한국어
What is dshfind?
dshfind is a community site built around DeepSeek Harness (DSH):
- 📖 Learn DSH principles — structured lessons from the basics up to a chapter-by-chapter deep dive of the Cordis paper: monads, coeffects, revertible effects, effect composition, spatiotemporal composability, and more.
- 🧩 Plugin marketplace — a live index of DSH plugins, automatically aggregated from the GitHub topic
dsh-plugin.
- 📚 Docs & community — the official DSH docs translated into four languages, plus a forum and per-plugin discussions.
Tech stack
- Astro 7 + Preact islands (
apps/web)
- Site i18n in four languages (English, 简体中文, 日本語 and 한국어 UI)
- MDX for lesson content (
src/content/lessons)
- Tailwind CSS · deployed on Cloudflare Workers + D1
Documentation
English
简体中文
Getting started
pnpm install
pnpm dev
Then open http://localhost:4321.
Regenerating plugin data
Plugins from the GitHub topic dsh-plugin are synced daily into the project database; the static snapshots (plugin library, homepage rails) are generated from it. To regenerate the snapshots only (needs database credentials in .env.local):
pnpm gen:data
To run a full refresh — GitHub sync, download counts, install methods, snapshots, build check — and commit only the generated files:
pnpm refresh
Submit your plugin
Add the dsh-plugin topic to your public GitHub repository. A scheduled job syncs the topic every day at 02:17 UTC, so a newly tagged repo shows up in the marketplace within about a day — no issue or PR needed.
If it still isn't listed after two syncs, open an issue: the repo may have been filtered as off-topic, or it may be private/archived.
Contributing
Issues and PRs are welcome:
Friend links
- DSH Desktop — a modern desktop app for the DeepSeek Harness (DSH) plugin ecosystem (GitHub)
- MZYAI GEO (妙智云) — an open-source GEO platform that gets your site accurately cited and recommended in AI answers from DeepSeek, Kimi, Doubao, GLM, Hunyuan and more (GitHub)
