LivXue/dsh-plugin-shop

The most comprehensive DeepSeek Harness plugin market — refreshed daily, sourced across the Internet, reviewed before publishing.

What it solves

dsh-plugin-shop is a plugin marketplace for the DeepSeek Harness (dsh), allowing users to discover, install, enable, and update plugins from a browsable, git-auditable catalog. It eliminates the need to manually search for compatible plugins and provides a mechanical filtering system to ensure basic quality and auditability.

How it works

The project consists of two main components: a daily build pipeline and a client-side interface.

  1. Harvesting: The system automatically scans npm and GitHub for packages and repositories using the dsh-plugin or deepseek-harness keywords.
  2. Gating: Every candidate is put through a five-step mechanical gate (checking for a valid bundle, license, installability, integrity, and description) to filter out low-quality or broken plugins.
  3. Cataloging: Valid plugins are categorized into one of seven categories and their peer dependencies are recorded.
  4. Client Interface: The plugin shop is integrated into the dsh Settings tab. It fetches a content-addressed JSON catalog from GitHub Pages or npm, verifies the SHA256 hash, and checks if the plugin's peer dependencies are met by the user's local profile before marking it as "Incompatible."

Who it’s for

Users of the DeepSeek Harness who want to extend its functionality via plugins without having to manually manage installations from arbitrary URLs.

Highlights

  • Automated Discovery: Automatically harvests plugins from the rest of the public registry without requiring a manual submission process.
  • Git-Auditable Catalog: Every change to the catalog is committed to git, making every addition or removal a reviewable diff.
  • Mechanical Filtering: Rejects candidates based on 17 distinct, author-readable reasons, ensuring a baseline of level of quality.
  • Zero-Privilege UI: The browser interface is decoupled from the runtime, ensuring that compromising the UI does not compromise the system runtime.
  • Dependency Validation: Performs local dependency checks against the user's specific profile to accurately report compatibility.

Related

  • Project
    yifanzhang-pro/recurrent-looped-tranformer

    A transformer architecture that introduces a recurrent feedback loop in the decoder to enable latent reasoning with infinite temporal depth as sequences grow.

  • Project
    kitfunso/hippo-memory

    A biologically-inspired memory layer for AI agents that uses decay and reinforcement to ensure agents remember important lessons and forget trivial details across different tools.

  • Project
    darkrishabh/agent-skills-eval

    A test runner for the Agent Skills ecosystem that empirically measures the performance lift of domain knowledge files by comparing model outputs with and without those skills.

  • Project
    mfish-qf/mfish-nocode

    A low-code/no-code platform that integrates AI assistants and multi-tenant LLM management to help users quickly build enterprise business systems and dashboards.