Integrate WorkBuddy/CodeBuddy

Configure WorkBuddy/CodeBuddy with a Nusabase OpenAI-compatible model through models.json.

WorkBuddy/CodeBuddy can add custom models through a local models.json file. Nusabase Chat Completions can be used as an OpenAI-compatible model endpoint.

When To Use This

• You want a desktop or editor-based agent coding assistant.
• You need models.json to pin team model choices.
• You want to configure lite or reasoning model relationships for CodeBuddy.

Prerequisites

• WorkBuddy/CodeBuddy is installed and signed in.
• You have a Nusabase API key: sk-....
• Your machine can reach https://api.nusabase.io.
• The example model qwen3.5-flash appears in the current pricing docs. Use /v1/models as the source of truth.

Config File Location

User-level config:

~/.codebuddy/models.json

On Windows this is usually:

C:\Users\<your-user>\.codebuddy\models.json

Project-level config:

<project-root>\.codebuddy\models.json

Project-level availableModels overrides the user-level list instead of merging with it. Do not commit project configs that contain a real API key.

Write models.json

The CodeBuddy official example expects apiKey to contain the actual key value, not an environment variable name.

{
  "models": [
    {
      "id": "nusabase-qwen3.5-flash",
      "name": "Nusabase qwen3.5-flash",
      "vendor": "Nusabase",
      "url": "https://api.nusabase.io/v1/chat/completions",
      "apiKey": "sk-your_api_key",
      "maxInputTokens": 128000,
      "maxOutputTokens": 8192,
      "supportsToolCall": true,
      "supportsImages": false,
      "relatedModels": {
        "lite": "nusabase-qwen3.5-flash",
        "reasoning": "nusabase-qwen3.5-flash"
      }
    }
  ],
  "availableModels": [
    "nusabase-qwen3.5-flash"
  ]
}

Field notes:

Field	Description
id	Internal CodeBuddy model ID. Use an nusabase- prefix.
name	Display name in the model picker.
vendor	Display vendor, use Nusabase.
url	Full Chat Completions URL, ending in /v1/chat/completions.
apiKey	Actual API key value.
supportsToolCall	Set to true only when the selected model/channel supports tool calls.
supportsImages	Usually false for text models; configure according to real model capability.

Restart And Select Model

1. Quit WorkBuddy/CodeBuddy completely.
2. Reopen the app so it reads the new models.json.
3. Select Nusabase qwen3.5-flash in the model picker.

Verify API

curl https://api.nusabase.io/v1/chat/completions \
  -H "Authorization: Bearer sk-your_api_key" \
  -H "Content-Type: application/json" \
  -d '{"model":"qwen3.5-flash","messages":[{"role":"user","content":"hi"}],"stream":false}'

Troubleshooting

Problem	Likely Cause	Fix
Model picker does not show the model	Wrong file location or app was not fully restarted	Confirm .codebuddy/models.json, then quit and reopen the app.
401	apiKey is wrong or expired	Verify the same key with curl first.
404	url is not the full Chat Completions URL	Use https://api.nusabase.io/v1/chat/completions.
Project config overrides user config	Project-level availableModels does not merge with user-level config	List all required model IDs in project config.

Use Models and Pricing as the source of truth for model names, context length, and price.

References

• CodeBuddy models.json guide
• CodeBuddy CLI Models
• First Request Example
• Models and Pricing
• Error Codes