This article introduces the DeepSeek Harness plugin dsh-dough-mix. When developing domain-specific agents, it is often necessary to integrate concrete business logic. For example, for food-related agents, quickly generating dough mixing ratios and noodle-making recipes is a common functional requirement. As a pure Node.js plugin, dsh-dough-mix provides developers with this capability.

Plugin Positioning

This is a network-enabled tool plugin maintained by uckkk. Its core value lies in returning noodle-making recipes based on parameters, enabling agents to execute dough mixing ratio calculations locally without relying on external network services.

Core Features

The plugin registers a tool named dough_mix. Its function is: Dough mixing ratio — returns noodle-making recipes based on parameters.

From a technical implementation standpoint, it is a pure Node.js implementation with no network dependencies. The following runtime dependencies must be met:

  • @deepseek-ai/cordis: Version must satisfy ^4.0.1.
  • @deepseek-ai/dsh-tools: Version must satisfy >=0.1.0-rc.6.

Installation and Activation

In a DSH environment, install the plugin using the following command:

dsh plugin add github:uckkk/dsh-dough-mix

Once installed, the plugin becomes active in the current session, allowing you to invoke its registered tools.

Typical Usage

After installation, simply invoke the tools registered by this plugin during a session. The agent will generate the corresponding dough mixing ratios and noodle-making recipes based on the input parameters, using pure Node.js logic.

Applicable Scenarios and Precautions

This plugin is well-suited for agent scenarios that require integrating recipe or cooking logic. Before use, it is recommended to review the source code and license (MIT). Since the plugin runs with the permissions of the current DSH process, please ensure you are aware of the security implications of its code logic.

Conclusion

This plugin offers a way to obtain noodle-making recipes without making network requests, making it ideal for integration into locally running agents. For more details, refer to the Catalog or the GitHub repository.