Skip to content
  • Home
  • Plugin manual
  • deepseek-harness-model-config

MarvekG/deepseek-harness-model-config

110Last commit Aug 14, 2026

deepseek-harness-model-config DSH plugin

This plugin extends the DeepSeek Harness Web UI with a dedicated settings page for advanced model configuration. It allows you to create custom model endpoints by specifying name, URL, API key, and protocol (openai-completions, openai-responses, anthropic-messages). The plugin also fetches available models and auto-completes their context window, output limits, input modalities, and reasoning capabilities via models.dev.

ui-productivitydeveloper-tools

How to install the deepseek-harness-model-config DSH plugin

dsh plugin --profile web add github:MarvekG/deepseek-harness-model-config

Copying does not run this command. Review the repository and version before installing the deepseek-harness-model-config DSH plugin.

deepseek-harness-model-config DSH plugin data source

deepseek-harness-model-config DSH plugin snapshot date: Aug 16, 2026

discovered

What the deepseek-harness-model-config DSH plugin can do

  • Create custom model endpoints with name, URL, API key, and protocol.
  • Support for openai-completions, openai-responses, and anthropic-messages protocols.
  • Fetch candidate models via the unified GET /models endpoint and select/deselect all.
  • Auto-complete missing context window, max output, input modalities, and reasoning ability from models.dev.
  • Edit each model's capacity, text/image input, and reasoningEfforts before saving.
  • View configuration preview without exposing API keys; keys are stored via Harness credential storage.

Where the deepseek-harness-model-config DSH plugin fits

  • Add a custom OpenAI-compatible endpoint to the Harness Web UI.
  • Configure Anthropic Claude models with specific reasoning efforts.
  • Batch-select or deselect models from a provider and fine-tune their parameters.
  • Preview the final model configuration without exposing sensitive API keys.

Who the deepseek-harness-model-config DSH plugin is for

  • DeepSeek Harness Web UI users who need to manage multiple model providers.
  • Developers integrating custom LLM endpoints into the Harness ecosystem.

deepseek-harness-model-config DSH plugin limitations

  • Only works within the DeepSeek Harness Web UI profile (requires running dsh web).
  • Supports only three protocols: openai-completions, openai-responses, anthropic-messages.
  • Model auto-completion relies on the models.dev service; availability is not guaranteed.
  • API keys must be provided by the user; there is no built-in key generation.

deepseek-harness-model-config DSH plugin: from the repository README

Quoted from the MarvekG/deepseek-harness-model-config README, the upstream source of the deepseek-harness-model-config DSH plugin. Copyright remains with the original authors.

为 [DeepSeek Harness](https://github.com/deepseek-harness/deepseek-harness) Web UI 添加“模型高级配置”页面，用于创建自定义模型端点并配置每个模型的能力参数。 ## 功能 - 新增自定义端点：名称、URL、API Key 和协议。 - 支持 `openai-completions`、`openai-responses`、`anthropic-messages`。 - 通过统一的 `GET /models` 流程获取候选模型，支持全选、反选和全不选。 - 从 `models.dev` 补全缺失的上下文窗口、最大输出、输入模态和推理能力。 - 在保存前编辑每个模型的容量、文本/图片输入和 `reasoningEfforts`，并查看不含密钥的配置预览。 - 所有界面文案支持中文和英文，跟随 Harness 语言设置。 - API Key 仅经 Harness 凭据存储写入，不会进入 `settings.yaml` 或配置预览。 打开设置： <img width="1613" height="1011" alt="image" src="https://github.com/user-attachments/assets/132b604b-b5ed-4986-87eb-c307d27cd8ba" /> 新增端点： <img width="1180" height="892" alt="image" src="https://github.com/user-attachments/assets/209bd3d6-ae39-489e-ae15-eb215c6c1126" /> 获取模型，并自动获取推理强度、上下文大小、输出强度等参数： <img width="1610" height="1439" alt="image" src="https://github.com/user-attachments/assets/4511b643-2d09-4a86-bb7f-5c4f7b6620a7" /> ## 安装 需要已安装并可运行的 `dsh`。默认从 GitHub 安装： ```sh dsh plugin --profile web add github:MarvekG/deepseek-harness-model-config dsh web ``` 打开 Web UI 的“设置 → 模型高级配置”，即可新增端点或编辑现有 `llm-pi-ai` 模型配置。 如需固定版本，在仓库地址后附加 commit SHA，例如 `github:MarvekG/deepseek-harness-model-config#<sha>`。 ## 本地调试 克隆本仓库后，在仓库根目

Read the full READMERepository license: MIT

deepseek-harness-model-config DSH plugin questions

How do I install this plugin?

Run `dsh plugin --profile web add github:MarvekG/deepseek-harness-model-config` from your terminal, then start the Web UI with `dsh web`. The plugin will appear under Settings > Model Advanced Configuration.

How do I update the plugin to a newer version?

First remove the old version with `dsh plugin --profile web remove dsh-models-config-plugin`, then reinstall using the same add command. Restart `dsh web` to apply the update.

Which protocols does the plugin support?

It supports three protocols: openai-completions, openai-responses, and anthropic-messages. These cover the most common API formats for OpenAI-compatible and Anthropic-compatible endpoints.

Is my API key safe when using this plugin?

Yes, the plugin stores API keys through Harness's credential storage, so they are not written to settings.yaml or exposed in the configuration preview. Your keys remain secure.

Can I edit the models after adding them?

Yes, before saving you can edit each model's capacity, text/image input, and reasoningEfforts. You can also view the full configuration preview without keys before finalizing.