Preface

When conducting conversations or developing agents in DeepSeek Harness (DSH), the model source typically requires separate configuration of API Keys, endpoints, or providers. If you have the Tencent WorkBuddy desktop app installed and logged in locally, the built-in models like GLM, DeepSeek, and Kimi within the app do not automatically appear in DSH’s model selector. The accounts and credentials on both sides are also maintained separately.

The DSH WorkBuddy Connect (maintained by corrinehu, GitHub repository corrinehu/dsh-workbuddy-connect) introduced below is designed for this scenario: After installing and enabling the plugin, it reuses the login state of the WorkBuddy desktop app, registers its included models into DSH, and makes them available for use by switching in the conversation window. The README describes it as “zero-configuration.” The plugin is categorized under model inference, currently at version 0.2.3, with an MIT license; its SkillHub directory page can be found at skillhub.cn/plugins/corrinehu/dsh-workbuddy-connect.

What is This

DSH WorkBuddy Connect is a DSH plugin designed to automatically integrate models included in the WorkBuddy desktop app (the README lists examples such as GLM-5.3, GLM-5.2, DeepSeek-V4-Pro, DeepSeek-V4-Flash, Kimi-K3, MiniMax-M3, Hy3, etc.) into DeepSeek Harness.

Compared to manually setting up a proxy API or entering keys model by model, this plugin relies on the app’s local login state. Account switches are synchronized with the WorkBuddy client, eliminating the need to repeatedly configure access credentials for the same set of models within DSH.

Core Features

Out-of-the-Box Experience

After installation and activation, WorkBuddy models will appear in the DSH model selector. The README states that no additional configuration is needed to use them in DSH conversations.

Account and Credit Viewing

On the web interface: Open Settings → Plugins → DSH WorkBuddy Connect card, expand it to view account information, token validity period, and remaining credits.

In the TUI interface, you can configure the authFile in /settings (as per the README).

Command-Line Status Check

The plugin provides a CLI for checking login status, diagnosing issues, and clearing credentials:

dsh plugin --profile <web|desktop|dsh-tui> exec dsh-workbuddy-connect status

Adding --json outputs a machine-readable format; there are also doctor for diagnostics and logout for clearing credentials (as per the README).

Installation and Activation

Prerequisites

  1. WorkBuddy desktop app must be installed and logged in. The plugin reuses the app’s login state, and account switches are automatically followed.
  2. Choose the corresponding installation command based on the DSH profile you are using. The plugin can run in Web, Desktop, and TUI interfaces.

The Web profile includes pre-built artifacts, and the README recommends this method:

dsh plugin --profile web add dsh-workbuddy-connect
dsh web

You can also install the Web version from the GitHub source:

dsh plugin --profile web add github:corrinehu/dsh-workbuddy-connect
dsh web

Desktop

dsh plugin --profile desktop add dsh-workbuddy-connect
dsh --profile desktop

TUI

dsh plugin --profile dsh-tui add dsh-workbuddy-connect
dsh --profile dsh-tui

The README notes: The dsh-tui profile requires pnpm 11 for installation; if your PATH contains another version, you might encounter ERR_PNPM_UNEXPECTED_STORE. You can use npx pnpm@11 as a workaround. This has been verified with dsh 0.1.1-rc.2.

After installation, switch to a WorkBuddy model in the model selector of the corresponding interface to start using it.

Typical Usage

After following the installation steps above, daily usage is similar to selecting a built-in provider in DSH:

  1. Ensure the WorkBuddy desktop app is logged in.
  2. Start DSH with the corresponding profile (dsh web, dsh --profile desktop, or dsh --profile dsh-tui).
  3. Select a model provided by WorkBuddy in the model selector to begin a conversation.
  4. When you need to check credits or token validity: On the web interface, open the settings card; or execute the status command:
dsh plugin --profile web exec dsh-workbuddy-connect status

Applicable Scenarios and Notes

Who Is It For

  • Users who already have the WorkBuddy desktop app installed locally and wish to directly use the models included in the app within the same DSH conversation/agent workflow, without building an additional API layer.
  • Developers who need to uniformly select the WorkBuddy model pipeline in any of the DSH Web, Desktop, or TUI interfaces.

Known Limitations

The README states:

  • Verified on macOS with DSH Web / Desktop / TUI profiles (0.1.1-rc.2+, Node 22+). The default credential paths on Windows / Linux have not been verified; if necessary, you can specify the actual location via the environment variable WORKBUDDY_AUTH_FILE.
  • Depends on the WorkBuddy client interface (not an official open API); the plugin may need to be adjusted following updates to WorkBuddy.

Before Installation, Please Note

The DSH ecosystem philosophy is “everything is a plugin”; SkillHub and similar platforms are community directories and have no official affiliation with DeepSeek or High-Flyer. The plugin runs with the permissions of the current dsh process and reads the local WorkBuddy-related credential paths. Before installation, you should read the GitHub source code and the MIT license, and be aware of the disclaimer in the README: The project is for personal learning and research use only, must comply with WorkBuddy’s terms of service, and is not affiliated with Tencent, WorkBuddy, or DeepSeek.

Links