Introduction

In use cases of DeepSeek Harness (DSH), the model catalog and reasoning tiers for the llm-pi-ai route often need to be updated in response to endpoint changes. When an endpoint provides new models or capabilities (such as contextWindow, maxTokens, input, etc.), this information must be synchronized into the system configuration. At the same time, the “Get Available Models” button may malfunction in some environments. The tool introduced in this article performs this synchronization through official interfaces and does not require any additional patching.

What Is This

This is a DSH plugin maintained by CN-WenYu, primarily used to maintain the model catalog and reasoning tiers for the llm-pi-ai route. It solves the following problems: when an endpoint provides new information such as models, contextWindow, maxTokens, or input, how to automatically complete and update it; and how to fix the functionality of the “Get Available Models” button.

Core Features

  • Synchronize models and capabilities: When an endpoint returns new models or capabilities, the plugin automatically completes this information.
  • Synchronize reasoning tiers: Reasoning information provided by the endpoint is written into the thinking tiers.
  • Fix interactive buttons: Fixes both the display and functionality of the “Get Available Models” button.
  • Use official interfaces with security in mind: The plugin operates only through official settings seams, without applying patches, and supports safe uninstallation.

Installation and Enablement

Installation must be performed through the official plugin directory. Please visit the GitHub repository to obtain the source code or installation link, and then perform the installation in the DSH management interface.

Typical Usage

After the plugin is enabled, no manual intervention is required. During interactions between DSH and the endpoint, the plugin automatically detects the model list and capabilities returned by the endpoint, and updates the local model catalog accordingly. If the endpoint returns reasoning information, the thinking tiers are written automatically.

Applicable Scenarios and Notes

  • Applicable scenario: DSH users who need to dynamically manage the model catalog and reasoning tiers for the llm-pi-ai route.
  • Notes: The plugin runs with the permissions of the current DSH process. Since the license field is empty, you should review the source code before installation to ensure it meets your security or compliance requirements.

Conclusion

This plugin solves the issues of model catalog synchronization and button repair through official interfaces, providing a patch-free and uninstallable maintenance solution. For details, visit the SkillHub directory or the GitHub repository.