In the development practice of DeepSeek Harness (DSH), attaching domain-specific tools to an agent is a common way to extend its capabilities. When handling physical parameter queries related to garment care, relying directly on a general-purpose large model is often insufficiently accurate or lacks concrete numerical references. The dsh-iron-temp plugin is built to address this need, turning garment ironing temperature queries into a deterministic tool function.

Plugin Positioning

This is a DSH plugin for querying garment ironing temperatures, maintained by developer uckkk. It takes parameters as input and returns clear temperature recommendations, solving the problem of agents lacking concrete parameter support when handling garment care.

Core Function

The plugin registers a tool named iron_temp. When using it, pass parameters such as the garment material, and the tool returns a clear ironing temperature recommendation.

Installation and Enablement

  1. Run the following command to install the plugin:
dsh plugin add github:uckkk/dsh-iron-temp
  1. After installation, the tool registered by the plugin can be invoked in the session.

Typical Usage

After installation, directly invoke the iron_temp tool in a DSH session. The tool accepts parameters and returns the recommended temperature.

Applicable Scenarios and Notes

  • Applicable scenarios: Suitable for scenarios where the agent needs to handle garment care, parameter queries, and similar tasks.
  • Technical details: The plugin is implemented in pure Node.js, has no runtime network dependencies, and has a small performance overhead.
  • Prerequisites: Ensure the current environment has @deepseek-ai/cordis (^4.0.1) and @deepseek-ai/dsh-tools (>=0.1.0-rc.6) installed.
  • Notes: Before installation, it is recommended to review its source code implementation and MIT license to ensure compliance with project security standards.

Summary

By providing a standardized temperature query interface, the plugin fills the gap for domain-specific tools. To view the source code or more details, you can visit its GitHub repository or skill directory.