To connect Tencent WorkBuddy to an OpenAI-compatible Chat Completions service, add it as a custom API model, then select that model in the conversation. The important fields are the provider’s API URL, its own API key, and the exact model ID accepted by that endpoint.
This is the Tencent WorkBuddy client setup. Feishu’s Doubao Work Partner is a different product with a separate model-settings flow and function-calling requirement; use its official model-management guide instead.
The official WorkBuddy guide specifies an OpenAI-compatible Chat Completions API. A successful text reply does not prove that the endpoint supports WorkBuddy’s tool-using tasks.
Add a custom API model
- In WorkBuddy, open Avatar → System Settings → Model, then add a model.
- Choose Custom API or Custom when the service is not in the provider list.
- Enter the API URL, the API key issued by that service, and the exact model ID from its documentation or account.
- Use Test connection, save the model, then select it in the conversation model picker. Saving does not switch every existing chat automatically.
WorkBuddy’s documentation says user model settings are stored locally, by default in ~/.workbuddy-ai/models.json or %USERPROFILE%\.workbuddy-ai\models.json on Windows. Treat that file as sensitive and do not upload it to a public repository or shared troubleshooting post.
Use ChinaGenAI for a short text chat
ChinaGenAI exposes a generic OpenAI-compatible Chat Completions route. These values select the site’s automatic route; free is a client label and does not name or guarantee a particular upstream model.
| WorkBuddy field | Value |
|---|---|
| Provider | Custom API |
| API URL | https://chinagenai.com/v1 |
| API key | Your ChinaGenAI site key, entered in WorkBuddy on your device |
| Model ID | free |
| Full Chat Completions URL, if requested | https://chinagenai.com/v1/chat/completions |
WorkBuddy’s default custom-protocol setting normalizes the URL to end in /chat/completions. Keep /v1 in the API URL. If you enable custom protocol, enter the complete URL the client should call, including /v1/chat/completions. This switch changes path handling; it does not convert another API format into Chat Completions.
Check service availability and site-key availability before connecting. ChinaGenAI keys are valid only for this site’s route; they are not OpenAI, DeepSeek or Volcengine Ark provider keys. For a direct Doubao API connection, see the Volcengine Ark setup.
Check limits and tool support
- Short output: the ChinaGenAI route currently caps generated output at 256 tokens. This can be too short for multi-step WorkBuddy tasks.
- Tool calls: WorkBuddy says many execution tasks need function calling. ChinaGenAI accepts and forwards standard function-tool fields, but the site’s upstream model is automatic and its WorkBuddy tool execution has not been verified. Begin with a short text chat; do not assume file, connector or other agent actions will work.
- Other API families: this route is Chat Completions only. It does not expose the Responses API, image input, audio, embeddings or files.
- Usage limits: site requests are subject to current per-key, network and site-wide limits. Read the live access and limit details before using it for a workflow.
Diagnose a failed connection
- 401: check that the API key came from the same service as the configured API URL. A ChinaGenAI site key and a provider key are not interchangeable.
- 404: compare the final request path. Keep the API version path such as
/v1; remove a duplicated/chat/completions, or use custom protocol only when you enter the full route. - Model not found: use the exact upstream model ID, including the provider’s required prefix or deployment name. A friendly display name may be a different field.
- Tool task fails after text chat works: verify that the selected model and endpoint support Chat Completions function calling, then enable WorkBuddy’s tool-call capability only for that service. Connection testing alone does not test every capability.
- Output ends early: inspect the response’s
finish_reasonand output-token limit. The ChinaGenAI route’s 256-token cap may truncate a longer task.
For a local check of a captured response, use the response checker. It does not send a model request.
Official references
WorkBuddy: Model Configuration
WorkBuddy’s model settings and endpoint behavior were checked against its official documentation on 4 November 2026. This is a written configuration guide; no WorkBuddy client session or model request was run.
Free API access and limits
Check the site free API access and limits before trying to obtain a site key. Read the current status and request limits before using a site key. It is not a source of official provider keys.
ChinaGenAI is not affiliated with the providers named here. This release contains setup guidance and explicitly labeled fictional examples, not claimed live model benchmarks.