In development scenarios for DeepSeek Harness (DSH), agents may sometimes make mistakes or need to handle negative interactions. At such times, generic apology responses often lack sincerity or specificity. Using a structured method to generate apology suggestions can help improve interaction quality.

dsh-apology is a DSH plugin that provides a set of apology formulas returning structured methods and suggestions based on the scenario. The plugin is maintained by uckkk and belongs to the model reasoning category. It aims to address issues with generation quality and structuring when agents handle apology scenarios.

Core Features

  • Apology Formula: Returns structured methods and suggestions based on the scenario
  • Technical Implementation: Pure Node.js implementation, no network dependencies

Installation and Activation

To install the plugin, run the following command:

dsh plugin add github:uckkk/dsh-apology

After installation is complete, you can invoke the tools registered by the plugin in a session.

Technical Details

The plugin runs under the permissions of the current DSH process. Please be sure to check the source code and license before use. Its dependencies are as follows:
* @deepseek-ai/cordis (^4.0.1)
* @deepseek-ai/dsh-tools (>=0.1.0-rc.6)

Summary

This plugin is suitable for scenarios that require improved interaction fault tolerance or handling of negative interactions.

  • Technical Catalog: https://www.skillhub.cn/plugins/uckkk/dsh-apology
  • Code Repository: https://github.com/uckkk/dsh-apology