Configuring WorkBuddy
WorkBuddy is Tencent's local AI workspace. Its custom-model support connects wylon chat models through the OpenAI Chat Completions-compatible API.
1. Add a model
- Open WorkBuddy and go to Settings.
- Under Features, select Models, then Add Model.
- Enter the values below and choose Test connection.
- After
Connection succeededappears, choose Save.
| Field | Value |
|---|---|
| Provider | Custom |
| Endpoint | https://api.wylon.cn/v1 |
| API Key | Your wylon API key |
| Model Name | A complete model ID, such as ZhipuAI/GLM-5.3 |
Endpoint and Custom Protocol
Keep Custom Protocol off and use https://api.wylon.cn/v1 so WorkBuddy appends /chat/completions. If you turn it on, use the full endpoint https://api.wylon.cn/v1/chat/completions.
info
Choose one URL modeDo not enter the full
/chat/completions path while automatic path completion is still enabled. The duplicated path returns 404.Model Name field
The model name is sent directly as the API model value. Copy the complete, case-sensitive ID from the Model Plaza or GET /v1/models.
2. Select and verify the model
- Return to the new-task screen and open the model selector.
- Select the model under Custom models.
- Send a simple message. A normal response with the selected model shown confirms the configuration.
3. Configuration file location
| System | Path |
|---|---|
| macOS | ~/.workbuddy-ai/models.json |
| Windows | .workbuddy-ai\models.json in your user directory |
WorkBuddy manages this file automatically. Prefer the settings UI and inspect the file only when troubleshooting.
warning
Do not commit the local model configurationWorkBuddy reads process-level variables from
process.env, but does not resolve the env field in settings.json. The apiKey is therefore stored in plain text. Never commit this file or leave it on a shared computer, and remove unused custom models promptly.4. Use another model
- Open the Model Plaza, or request
GET https://api.wylon.cn/v1/models. - Copy the complete API model ID.
- Add a model under Settings → Models, or edit an existing model.
- Keep the endpoint and API key, and replace only the model name.
info
Create a new task to verify the model after switching. For tool use, long-context, or structured-output workloads, rerun the relevant capability checks.
