Configure CodeBuddy IDE

CodeBuddy IDE is an AI agent and coding assistant. It adds custom models through a local model configuration file, so you can connect InOneAPI with the OpenAI-compatible Chat Completions API.

1. Before you start

  1. Install CodeBuddy IDE and sign in.
  2. Open a project directory at least once so the application creates its local configuration directory.
  3. Create an API key in the InOneAPI console and copy the model ID from the model detail page exactly as shown. YOUR_MODEL_ID is only a documentation placeholder.

The agent tools entry on the model detail page can generate the complete configuration content with the key and the model ID already filled in.

2. Configure the local model

Create or edit the user-level configuration file:

  • macOS / Linux: ~/.codebuddy/models.json
  • Windows: C:\Users\<your-username>\.codebuddy\models.json

To limit the configuration to one project, create a project-level file at <your-project>/.codebuddy/models.json instead.

Write the following configuration:

{
  "models": [
    {
      "id": "YOUR_MODEL_ID",
      "name": "YOUR_MODEL_ID",
      "vendor": "InOneAPI",
      "url": "https://api.inoneapi.com/v1/chat/completions",
      "apiKey": "ioa_sk_...",
      "supportsToolCall": true,
      "supportsImages": false
    }
  ],
  "availableModels": [
    "YOUR_MODEL_ID"
  ]
}

Keep the other entries when you merge the models array. availableModels is a display allowlist: when the field already exists, append the current ID instead of overwriting the list.

Field notes:

  • url is the full Chat Completions endpoint and must include /v1/chat/completions.
  • id must be exactly the model ID from the model detail page, not only its display name.
  • maxInputTokens and maxOutputTokens are optional. Set them only when you know the model's real limits; the page no longer guesses 128K / 8192.
  • supportsToolCall and supportsImages must match the model's real capabilities; change them to false when unsupported.

Save models.json as UTF-8 without a BOM. Some desktop builds fail to read the local model configuration when the JSON file starts with a UTF-8 BOM.

Set apiKey to the actual key value as the official field documentation describes. Substitutions such as ${INONEAPI_API_KEY} are not a documented, supported form.

3. Save and select the model

  1. Save the file. Official documentation supports hot reload with about one second of debouncing; restart only if the configuration does not apply.
  2. Select the model you just added in the model picker.

4. Optional: verify the API key

On macOS / Linux you can check in a terminal that the key and the model ID work:

curl https://api.inoneapi.com/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer ioa_sk_..." \
  -d '{"model":"YOUR_MODEL_ID","messages":[{"role":"user","content":"hi"}],"stream":false}'

A successful request means both the API key and the model ID are usable.

Common issues

  • Authentication Fails or 401: check that apiKey holds a real InOneAPI API key, and do not put the endpoint URL in the API key field.
  • Model not found or 404: check that id equals the model ID on the model detail page exactly.
  • The local model configuration cannot be read: check that models.json is valid JSON and saved as UTF-8 without a BOM.
  • The model does not appear in the picker: restart CodeBuddy IDE completely and confirm the file is at .codebuddy/models.json.

For production use, create a dedicated sub-key for CodeBuddy IDE and set its budget, model restrictions and alerts. See Team keys and usage.

This page covers CodeBuddy IDE only, and does not apply to the CodeBuddy CLI or any plugin builds. For WorkBuddy use the visual configuration guide.

Official reference: CodeBuddy IDE models.json.

CodeBuddy IDE · Documentation · InOneAPI