Connect an agent

WorkBuddy

WorkBuddy is a desktop AI work assistant with custom API support. Enter the Yomi API address, key and model ID to use the platform's chat models in WorkBuddy.

Prerequisites

Download the installer for your operating system from the WorkBuddy website and install it.
Sign in to the Yomi API console, create a key under API Keys and save it. It is visible only at creation.
Ensure your account has enough credit. If needed, add credit through the User Center's Add Credit page.

Configure Yomi API

Open WorkBuddy Settings, go to Models, click Add Model and choose Custom as the provider. The dialog notes that only OpenAI-compatible APIs are supported. Enter the following three fields and save:

WorkBuddy Add Model entry — Settings → Models → Add Model. Choose Custom at the bottom of the provider list. The screenshot uses the Chinese interface.

Endpoint address: https://api.yomiapi.com/v1/chat/completions (WorkBuddy requires the complete chat endpoint address)
API Key: YOUR_API_KEY (replace with your key)
Model name: Enter a model ID, such as claude-sonnet-5, gpt-5.6-sol; for lightweight everyday tasks, use deepseek-v4-flash

Use the model ID from the Model Catalog (such as claude-sonnet-5). A display name such as Claude Sonnet 5 causes calls to fail. See Models in the console for the complete catalog.

WorkBuddy custom model form — Add Model form: the endpoint address, API Key and model name fields. The screenshot uses the Chinese interface; fill these three fields with the values above.

Start and verify

Return to the main window, start a conversation, select your configured model and send a test message. A normal reply confirms successful setup. You can also find the call under Usage in the Yomi API console User Center.

Common questions

Connection failure or 404 response

Check that the endpoint address is https://api.yomiapi.com/v1/chat/completions. Missing or duplicated path segments produce an incorrect request path.

401 authentication failure

The key is incorrect or has been deleted. Check it in API Keys, then copy and paste it again without leading or trailing spaces.

Model does not exist

Use the model ID, not its display name. If the ID is correct, check that your key's group includes that model; the Model Catalog lists available models for each group.