WorkBuddy

The CodeBuddy IDE AI coding assistant — connect Models Hub's OpenAI-compatible models via a local models.json.

Introduction

WorkBuddy (CodeBuddy IDE) supports custom models through a local models.json file and is compatible with the OpenAI interface format, so it connects to Models Hub models seamlessly.

📦 Before you start

WorkBuddy / CodeBuddy IDE installed
Models Hub base URL: https://modelsok.com/v1 (the endpoint must use the full path, see below)
A valid Models Hub API key (generated in the console)
The model name you want to use (e.g. claude-sonnet-4-6, gemini-2.5-flash), which must exactly match the model ID in the Models Hub console

🗂️ Configuration file location

WorkBuddy manages custom models with a two-level models.json:

Level	Path	Notes
User	`~/.codebuddy/models.json`	Applies globally
Project	`<project-root>/.codebuddy/models.json`	Project-specific, takes precedence

The availableModels field in the project-level config completely overrides the user-level one.

🚀 Step 1: Edit models.json

Open (or create) the relevant models.json and add the Models Hub models:

{
  "models": [
    {
      "id": "claude-sonnet-4-6",
      "name": "Claude Sonnet 4.6 (Models Hub)",
      "vendor": "Models Hub",
      "url": "https://modelsok.com/v1/chat/completions",
      "apiKey": "ms_live_your_api_key",
      "maxInputTokens": 200000,
      "maxOutputTokens": 8192,
      "supportsToolCall": true,
      "supportsImages": true,
      "supportsReasoning": true
    },
    {
      "id": "gemini-2.5-flash",
      "name": "Gemini 2.5 Flash (Models Hub)",
      "vendor": "Models Hub",
      "url": "https://modelsok.com/v1/chat/completions",
      "apiKey": "ms_live_your_api_key",
      "maxInputTokens": 1000000,
      "maxOutputTokens": 8192,
      "supportsToolCall": true,
      "supportsImages": false
    }
  ],
  "availableModels": [
    "claude-sonnet-4-6",
    "gemini-2.5-flash"
  ]
}

Field reference

Field	Required	Description
`id`	Yes	Unique model identifier; must match the model ID in the Models Hub console
`name`	Yes	Display name shown in the IDE
`url`	Yes	The full Chat Completions endpoint, ending with `/chat/completions`
`apiKey`	Yes	The actual API key value (not an environment variable name)
`vendor`	Yes	Provider name; you can use `Models Hub`
`maxInputTokens` / `maxOutputTokens`	No	Maximum input/output tokens for the model
`supportsToolCall`	No	Whether tool/function calling is supported
`supportsImages`	No	Whether image input is supported
`supportsReasoning`	No	Whether reasoning mode is supported
`availableModels`	No	Restricts which models appear in the dropdown; show all if empty

WorkBuddy auto-reloads about 1 second after the config changes (debounced) — no restart needed. Select the Models Hub model you just added (e.g. claude-sonnet-4-6) in the model picker and send a test message:

Write a quicksort implementation in TypeScript

A normal response means the integration works. Custom models are automatically tagged custom for easy filtering in the UI.

❓ FAQ

Problem	Solution
New model doesn't appear in the picker	Check that `availableModels` includes the model ID and that the config has reloaded
401 / invalid API key	Make sure `apiKey` is the actual key value, not an environment variable name
`url must end with /chat/completions`	Make sure `url` is the full endpoint, not just `/v1`
Model not found	Make sure `id` exactly matches the model ID in the Models Hub console
Project config not taking effect	The project-level `models.json` overrides the user level — check `.codebuddy/models.json` in the project root
Output is truncated	Increase `maxOutputTokens`

📦 Before you start

🗂️ Configuration file location

🚀 Step 1: Edit models.json

✅ Step 2: Verify

❓ FAQ

On this page