When to Use
WorkBuddy supports configuring custom models. You can integrate Tokener CA as an OpenAI-compatible model provider so WorkBuddy can use platform models for code understanding, editing, explanation, and Agent tasks.
You only need four values to integrate:
| Setting | Value |
|---|---|
| API Key | Consumer API Key, for example stc-... |
| Endpoint | https://a.skypool.xyz/v1/chat/completions |
| Model ID | Platform model name, for example gemma4:26b |
| Model capabilities | Choose text, tool calling, image, and other capabilities based on the model's actual abilities |
Option 1: Configure in the WorkBuddy UI
If your version of WorkBuddy offers a "custom model" or "custom provider" UI, we recommend using it first.
Fill in the values below:
| Field | Value |
|---|---|
| Provider name | Tokener CA |
| Model name | gemma4:26b, or the platform model ID you want to use |
| API Key | Consumer API Key |
| Endpoint | https://a.skypool.xyz/v1/chat/completions |
| Protocol type | OpenAI-compatible |
After saving, restart WorkBuddy or refresh the model list, then select the model you just added in the model picker.
Option 2: Use a Local Configuration File
If your version supports local JSON configuration, you can add a Tokener CA model at the user level or project level.
Common configuration locations:
| Scope | Windows | macOS / Linux |
|---|---|---|
| User level | C:\Users\<username>\.workbuddy\models.json | ~/.workbuddy/models.json |
| Project level | <project_root>\.workbuddy\models.json | <project_root>/.workbuddy/models.json |
Some WorkBuddy versions reuse the CodeBuddy configuration directory. If .workbuddy doesn't take effect, check the sibling .codebuddy/models.json or .codebuddy/models.json in your user directory.
Example configuration:
{ "models": [ { "id": "gemma4:26b", "name": "Tokener CA gemma4 26B", "vendor": "Tokener CA", "url": "https://a.skypool.xyz/v1/chat/completions", "apiKey": "${SKYPOOL_API_KEY}", "maxInputTokens": 128000, "maxOutputTokens": 8192, "supportsToolCall": true, "supportsImages": false } ], "availableModels": [ "gemma4:26b" ]}If the model supports image input, set supportsImages to true. If you only do plain text and code tasks, keeping it false is safer.
Set Environment Variables
We recommend putting the API Key in an environment variable and having the configuration file only reference the variable.
Current PowerShell window:
$env:SKYPOOL_API_KEY="stc-..."Persist on Windows:
setx SKYPOOL_API_KEY "stc-..."macOS / Linux:
export SKYPOOL_API_KEY="stc-..."After setting it, reopen WorkBuddy to make sure it can read the new environment variable.
Verify the API
Before integrating with WorkBuddy, run a request with the same key and model to make sure it works:
curl -X POST "https://a.skypool.xyz/v1/chat/completions" \ -H "Authorization: Bearer <consumer_api_key>" \ -H "Content-Type: application/json" \ --data-raw '{ "model": "gemma4:26b", "messages": [ { "role": "user", "content": "Return a short message confirming the integration works." } ], "max_tokens": 64, "stream": false }'If curl succeeds but WorkBuddy can't make the call, first check that WorkBuddy has the full endpoint filled in, not just https://a.skypool.xyz/v1.
Troubleshooting Checklist
| Symptom | How to Fix |
|---|---|
| Model doesn't appear in the list | Check that availableModels includes the matching id, then save and restart WorkBuddy |
| Authentication fails | Check whether the environment variable is read by the WorkBuddy process, or temporarily fill in the key directly to verify |
404 | Check that the model ID comes from the Tokener CA model list |
| Wrong request URL | WorkBuddy's url usually needs the full /v1/chat/completions address |
| Tool calling misbehaves | First confirm whether the model is suitable for tool calling, then check the supportsToolCall setting |