Hysen Labs
Open-source project
hikariming/dshfind avatar
hikariming

dshfind

DSH (DeepSeek Harness) 原理学习、插件市场与最佳实践 · Learn DSH principles, plugin marketplace & best practices

202 stars9 forksTypeScriptLicense varies
DEEP OPEN-SOURCE ANALYSIS

Dshfind is a learning and plugin community for DeepSeek Harness

Dshfind is a community website built around DeepSeek Harness (DSH). It offers structured lessons, a plugin marketplace, and best practice guides for plugin authors.

What dshfind provides

The site organizes material around DeepSeek Harness in three areas. The lessons section walks readers from basic concepts up to a chapter by chapter reading of the Cordis paper, covering monads, coeffects, revertible effects, effect composition, and spatiotemporal composability. The plugin marketplace is a live index of DSH plugins pulled automatically from the GitHub topic dsh-plugin. The best practices area adds plugin development guides, a glossary, and community rankings of authors and projects.

How the plugin index stays current

Plugin and ranking data are generated from the GitHub topic dsh-plugin and require the GitHub CLI to regenerate locally with the command pnpm gen:data. A scheduled job syncs the topic every day at 02:17 UTC, so a newly tagged public repository appears in the marketplace within about a day without needing an issue or pull request. Repositories that stay unlisted after two syncs may have been filtered as off topic or marked private or archived.

Technology and tooling

The project is built with Next.js 16 using the App Router and React 19. Internationalization for the English and Chinese interface is handled by next-intl, lesson content is written in MDX and stored under src/content/lessons, and styling uses Tailwind CSS. The site is deployed on Vercel. Documentation covers a Vercel plus Railway production deployment and a public API and query guide, with Chinese versions of both.

Editorial conclusion

Dshfind is developed in TypeScript and uses Next.js 16 with the App Router, with lesson material stored as MDX files under src/content/lessons.

DEEP OPEN-SOURCE ANALYSIS

Official sources

Community notes

Community notes