DeepSeek Harness (DSH) uses a plugin-based architecture, allowing developers to extend system capabilities by registering tools. When building domain-specific workflows or agents, lightweight local logic checks are sometimes needed. dsh-routine is such a plugin focused on facial skincare steps, and it can return results without requiring external network interfaces.

Plugin Overview

This is a workflow plugin maintained by uckkk and open-sourced under the MIT license. It addresses a specific need: when relevant facial skincare routines need to be generated or queried based on parameters, the required steps can be obtained without relying on external APIs.

Core Capabilities

The plugin registers a tool named routine. By passing parameters, this tool can return the corresponding facial skincare routine.

Technical Implementation

This plugin is implemented purely in Node.js and does not include any network request dependencies. This means it runs quickly and has low resource usage.

Before running it, make sure the environment satisfies the following dependency versions:

{
  "peerDependencies": {
    "@deepseek-ai/cordis": "^4.0.1",
    "@deepseek-ai/dsh-tools": ">=0.1.0-rc.6"
  }
}

Installation

Install it by running the following command in the DSH CLI:

dsh plugin add github:uckkk/dsh-routine

Usage

After installation, you can directly invoke the routine tool registered by the plugin in a DSH session or workflow.

Use Cases and Precautions

It is suitable as a stage in a workflow for handling skincare logic. Because the plugin runs within the DSH process, be sure to review the source code and license before use.

Summary

dsh-routine is a lightweight local logic plugin for generating skincare steps. For more details, refer to the Plugin Directory or the Source Repository.