Prerequisites
- WorkBuddy / CodeBuddy installed (codebuddy.cn)
- A ClearMaas API key starting with
sk-
Setup via models.json
WorkBuddy supports file-based model configuration. Create or editmodels.json:
User-level (all projects): ~/.codebuddy/models.json
Project-level (single project): .codebuddy/models.json
Project-level settings override user-level for matching model IDs.
Adding ClearMaas
| Field | Description |
|---|---|
url | Full path to the chat completions endpoint (must include /chat/completions) |
vendor | Must be "OpenAI" �?only OpenAI interface format is supported |
supportsToolCall | Enable tool/function calling support |
supportsImages | Enable image input (vision) |
Setup via UI
Since 2025, WorkBuddy supports graphical model configuration:- Open WorkBuddy settings
- Navigate to Models or AI Settings
- Click Add Custom Model or Add Provider
- Fill in:
| Field | Value |
|---|---|
| Provider | OpenAI Compatible |
| API Key | sk-... |
| Base URL | https://api.clearmaas.com/v1 |
| Model ID | e.g., anthropic/claude-sonnet-4.6 |
| Max Input Tokens | 200000 |
| Max Output Tokens | 8192 |
Model format
When specifying the model ID, use ClearMaas’sprovider/model-name format:
| Model ID | Description |
|---|---|
anthropic/claude-sonnet-4.6 | Claude Sonnet 4.6 |
anthropic/claude-opus-4.8 | Claude Opus 4.8 |
openai/gpt-4o-mini | GPT-4o Mini |
google/gemini-2.5-pro | Gemini 2.5 Pro |
deepseek/deepseek-chat | DeepSeek Chat |
grok/grok-4-fast-reasoning | Grok 4 Fast Reasoning |
Ollama and local models
WorkBuddy also supports local models via Ollama. If you have both Ollama and ClearMaas configured, WorkBuddy can route between them:- Ollama:
http://localhost:11434/v1/chat/completions - ClearMaas:
https://api.clearmaas.com/v1/chat/completions
Custom protocol for gateways
If your ClearMaas URL uses non-standard paths, enable Custom Protocol in the advanced settings to skip path validation.Verification
- After adding ClearMaas, select the model from the model picker
- Ask a coding question or run a task
- Check your ClearMaas dashboard to confirm the request was routed