Back to directory

dsh-plugin-shop

A daily-refreshed plugin marketplace for DeepSeek Harness that automatically discovers, screens and lists npm packages and GitHub repositories tagged as dsh plugins, with every catalog change tracked in Git.

dsh-plugin-shop

dsh-plugin-shop is a plugin shop for DeepSeek Harness (dsh) that lets a user browse, install, enable, disable and update plugins from inside dsh's Settings menu, or have an agent run the same steps non-interactively with a required --profile option. Plugin management is handled through dsh plugin, which in turn invokes pnpm.

Its catalog is built automatically: it collects npm packages carrying the dsh-plugin keyword and GitHub repositories carrying the dsh-plugin or deepseek-harness topic, so an author does not need to submit an application or join a review queue. Every candidate is screened automatically on each build against seventeen predefined rejection reasons, with a detailed explanation recorded for each rejected plugin, and the shop checks each plugin's declared peer dependencies and supported dsh versions or profiles against a user's own environment before marking it compatible or incompatible.

The catalog rebuilds daily, with every change committed to Git for review, and eligible new plugins appear the next morning.

More in Agent skills

See all 581
  • Provenance gate

    github.com

    A deterministic authorization gateway that sits between an AI agent and its tools, tracing where each parameter of a side-effect call came from, and blocking the call or asking a human when the source does not check out.

  • A markdown framework of system rules that restates Isaac Asimov's Four Laws of Robotics as a set of prioritized loss calculations for large-language-model agents to follow, published as a discovery file for agents and crawlers to read.

  • Vitruvius

    github.com

    An engineering research agent that runs a discover, read, synthesize, verify and review loop over technical questions, reading sources directly rather than inferring from titles, and publishing its own scored verifier benchmark, weaknesses included.