Skip to content

command-code.md

Latest commit

 

History

History
67 lines (50 loc) · 3.58 KB

command-code.md

File metadata and controls

67 lines (50 loc) · 3.58 KB
Raw

Command Code

Connect Command Code's Provider API to AIO Proxy with an API Provider. Create an API key in Command Code Studio and make it available as CMD_API_KEY to the process running AIO Proxy.

Command Code documents API access for GOAT, Pro, Max, Team, and Provider plans. Coding subscriptions charge API requests against their plan credits and model allowances; the Go plan does not include API access. Check the current plan documentation for your account's limits.

Configuration

Merge the relevant entries below into your configuration's providers object. Choose models your plan allows from the live catalog.

{
  "providers": {
    "command-code": {
      "kind": "api",
      "apiKey": "{{env.CMD_API_KEY}}",
      "models": ["deepseek/deepseek-v4-flash"],
      "endpoints": {
        "baseURL": "https://api.commandcode.ai/provider/v1",
        "protocol": ["openai-compatible"],
      },
    },
    "command-code-claude": {
      "kind": "api",
      "apiKey": "{{env.CMD_API_KEY}}",
      "models": ["claude-sonnet-4-6"],
      "endpoints": [
        {
          "baseURL": "https://api.commandcode.ai/provider/v1",
          "protocol": "anthropic",
          "auth": "bearer",
        },
      ],
    },
  },
}

The separate Provider entries are intentional: Command Code accepts Claude models at /provider/v1/messages and other models at /provider/v1/chat/completions. Advertising both protocols for the same model would allow a same-protocol request to reach the wrong upstream endpoint. Separate model lists let AIO Proxy select the appropriate Provider and convert the client's protocol when needed.

Use endpoints to retain /provider/v1. A legacy top-level protocol/baseURL pair uses only the URL's origin for same-protocol passthrough and would send these requests to the wrong path.

Validate and reload after saving:

aio-proxy config validate
aio-proxy reload

Point your client at AIO Proxy's local endpoint, normally http://127.0.0.1:9317/v1, and select one of the configured model IDs. If AIO Proxy requires a caller API key, configure that key in the client; CMD_API_KEY is the upstream credential used by the proxy.

Dashboard

Create an API Provider, select OpenAI Chat Completions, and enter https://api.commandcode.ai/provider/v1 as its address. Enter your Command Code API key and add the non-Claude models you want to expose.

For Claude, create a separate API Provider with Anthropic Messages and the same address. The default x-api-key authentication is supported by Command Code; the independent-address editor also allows selecting Bearer to match the file example above. Add only the Claude models to this Provider.

New single-protocol entries preserve the full address. An existing legacy protocol/baseURL configuration keeps its historical behavior; use the endpoints configuration above to migrate one that needs the /provider/v1 prefix.

Troubleshooting

  • 404 or a missing /provider prefix: check that the saved Provider uses endpoints and the complete address above.
  • 400 invalid_request_error: check the model's protocol and its Provider's model list.
  • 401: check the API key supplied to AIO Proxy and the upstream key configured for the Provider.
  • 403 upgrade_required: the official API is unavailable on the Go plan.

See Command Code's API reference for current upstream errors and account requirements.