dsh-model-search: Adding Searchable Filtering to the DeepSeek Harness Model Selector

Preface

When using the model selector in DeepSeek Harness, if you need to find a model by provider, model name, or model ID, searching through them one by one can be slow. dsh-model-search adds a searchable filter to the model selector, making it convenient to directly enter keywords to locate candidate models.

What This Is

a1073097082/dsh-model-search is a plugin for the DeepSeek Harness model selector, maintained by a1073097082, licensed under MIT. The problem it solves is: after opening the model selector, you can search and filter by provider name, model name, or model ID.

Core Features

The plugin provides the following capabilities:

  • After opening the model selector, you can search by provider name, model name, or model ID.
  • Search keywords are case-insensitive.
  • Spaces act as wildcards; for example, GPT 6 can match GPT-5.6 and GPT-6.
  • The plugin is a client-only overlay; it does not change model routing or credentials.

Installation and Activation

First, run the following installation command:

dsh plugin --profile web add github:a1073097082/dsh-model-search

After installation, refresh the Harness page, then open the model selector to verify that search filtering is active.

Typical Usage

You can use it in the following way:

  1. Open the model selector in DeepSeek Harness.
  2. Enter keywords in the search box by provider name, model name, or model ID.
  3. For example, entering GPT 6 can match GPT-5.6 and GPT-6.
  4. If you have just installed it, refresh the Harness page first, then perform the above steps.

Applicable Scenarios and Notes

It is suitable for users who need to quickly filter models, providers, or model IDs in DeepSeek Harness. The plugin runs with the permissions of the current dsh process; you should check the source code and license before installation. This plugin is licensed under MIT; it only acts as a client-only overlay and does not alter model routing or credentials.

Conclusion

The value of dsh-model-search lies in providing a lightweight search filtering entry for the Harness model selector, reducing the cost of manual lookup. The GitHub address is as follows:

https://github.com/a1073097082/dsh-model-search

羽毛球分组比赛记分
小程序二维码

欢迎使用《羽毛球分组比赛记分》微信小程序

Xiaoye