Continue
Add an OpenAI model entry in Continue that points at DurinDoor.
Use Continue's OpenAI-compatible model configuration with a running gateway, a DurinDoor key, and a model ID or combo.
Copy the dashboard guide
Open CLI Tools → Continue, choose the endpoint, key, and model, then copy the JSON snippet. Continue is a guide card. It does not write configuration files or report an automatic connection status.
The examples below use Continue's JSON models configuration. If your version uses a different format, translate the same endpoint, provider, key, and model fields using that version's configuration editor.
Manual config
Merge this object into the models array in the Continue config file:
{
"apiBase": "http://localhost:20128/v1",
"title": "provider/model-id",
"model": "provider/model-id",
"provider": "openai",
"apiKey": "YOUR_DURINDOOR_API_KEY"
}A full file with one entry looks like this:
{
"models": [
{
"apiBase": "http://localhost:20128/v1",
"title": "provider/model-id",
"model": "provider/model-id",
"provider": "openai",
"apiKey": "YOUR_DURINDOOR_API_KEY"
}
]
}Add more objects in that array to switch combos from Continue's model picker. A remote DurinDoor needs an https URL and a host Continue can reach.
provider must be "openai". apiBase must include /v1. A missing suffix is the usual "Continue cannot reach DurinDoor" miss.
Check it works
Reload Continue. Send a small prompt. Confirm the row on Usage.
| Symptom | Check |
|---|---|
| Cannot reach DurinDoor | apiBase, firewall, and whether the gateway is running on port 20128 |
| Invalid model | Use an id from GET /v1/models or a combo name |
| Provider parameter error | The selected upstream may not support a field Continue sends. Try another model or combo member |
| Tools or large context fail | Test the exact workflow before relying on a new combo |