CodeBuddy Code

Check CodeBuddy Code

Run:

codebuddy --version

If it is not installed yet, install it with the official package:

npm install -g @tencent-ai/codebuddy-code

Create models.json

CodeBuddy Code registers custom models through models.json. You can use either user-level or project-level configuration:

• User-level: ~/.codebuddy/models.json, applies to all projects
• Project-level: <project-root>/.codebuddy/models.json, applies only to the current project and has higher priority

The user-level path on Windows is usually:

%USERPROFILE%\.codebuddy\models.json

Write the following content:

{
  "models": [
    {
      "id": "deepseek-v4-pro",
      "name": "deepseek-v4-pro",
      "vendor": "YoloRouter",
      "url": "https://api.yolorouter.com/v1/chat/completions",
      "apiKey": "${YOLOROUTER_API_KEY}",
      "maxInputTokens": 128000,
      "maxOutputTokens": 8192,
      "supportsToolCall": true,
      "supportsImages": false
    },
    {
      "id": "deepseek-v4-flash",
      "name": "deepseek-v4-flash",
      "vendor": "YoloRouter",
      "url": "https://api.yolorouter.com/v1/chat/completions",
      "apiKey": "${YOLOROUTER_API_KEY}",
      "maxInputTokens": 128000,
      "maxOutputTokens": 8192,
      "supportsToolCall": true,
      "supportsImages": false
    }
  ],
  "availableModels": [
    "deepseek-v4-pro",
    "deepseek-v4-flash"
  ]
}

If your Yolo Router endpoint already includes /v1, for example https://api.yolorouter.com/v1, set url to https://api.yolorouter.com/v1/chat/completions; do not duplicate it as /v1/v1/chat/completions.

Set the Yolo Router API Key

The models.json above uses ${YOLOROUTER_API_KEY}, so set the environment variable before starting CodeBuddy Code.

export YOLOROUTER_API_KEY="sk-yolo-..."

$env:YOLOROUTER_API_KEY = "sk-yolo-..."

set YOLOROUTER_API_KEY=sk-yolo-...

Do not write the real API Key into models.json or commit it to your repository.

Start and select the model

Open any project directory and start CodeBuddy Code:

codebuddy --model deepseek-v4-pro

For a print-mode test, also pass --model explicitly:

codebuddy --model deepseek-v4-pro -p "Introduce the model you are using in one sentence"

After a normal response, check the Yolo Router console request log and confirm the request used deepseek-v4-pro.

Optional: set the sub-agent model

If you want CodeBuddy Code sub-agents to use the faster model, set:

export CODEBUDDY_CODE_SUBAGENT_MODEL="deepseek-v4-flash"

$env:CODEBUDDY_CODE_SUBAGENT_MODEL = "deepseek-v4-flash"

set CODEBUDDY_CODE_SUBAGENT_MODEL=deepseek-v4-flash