dsh-recommend

As DeepSeek Harness plugins grow in number, the real challenge shifts from finding any plugin to judging which ones are active, relevant, and worth trying. dsh-recommend addresses this by turning plugin discovery into a transparent, recomputable ranking: it continuously collects repositories tagged with the dsh-plugin topic, scores them using a public model that considers maintenance, popularity, quality, and ecosystem signals, and publishes the result as a shared data source.

The plugin is not just a leaderboard; it is designed to be auditable. The scoring formula, weights, and raw data are public, while exclusion lists, plugin-feature verification, and an automated data pipeline help reduce noise from placeholders, non-plugin repositories, and low-quality entries. Inside DSH, users can browse, search, filter, and refresh ranking data through a settings-page tab, while the static ranking site and the plugin consume the same source of truth.

dsh-recommend is useful for DSH users who want to discover and compare plugins more confidently, plugin authors who want to understand how their projects appear in the public ecosystem, and maintainers or developers who need consistent ranking data for tools, reports, or ecosystem analysis.