IDE agents
Connect Cursor, Cline, Roo Code, Kilo Code and Continue to Nodul Router as an OpenAI-compatible provider.
Coding agents that accept an OpenAI-compatible provider work with the router. You always need three values:
| Setting | Value |
|---|---|
| Base URL | https://router.nodul.ru/api/v1 |
| API key | Your Router key, sk-nr-… |
| Model | A model id from the catalog, for example openai/gpt-4o-mini |
Pick a model with the Tools capability: agents call tools on every step. Agents send long prompts, so watch the spend in the cabinet logs and set a budget on the key.
Cline, Roo Code, Kilo Code
- Open the extension's settings and choose the API provider OpenAI Compatible.
- Base URL:
https://router.nodul.ru/api/v1. - API key: your Router key.
- Model ID: the model id from the catalog.
- Set the context window and the output limit to the model's
context_lengthandmax_completion_tokensfromGET /api/v1/models.
Continue
Add a model to config.yaml:
models:
- name: Nodul Router
provider: openai
model: openai/gpt-4o-mini
apiBase: https://router.nodul.ru/api/v1
apiKey: sk-nr-...
roles: [chat, edit, apply]Cursor
- Open Cursor Settings → Models.
- In API Keys, enter the Router key as the OpenAI API key, turn on Override OpenAI Base URL and enter
https://router.nodul.ru/api/v1. - Add the model id from the catalog as a custom model and turn it on.
Cursor sends some requests through its own servers, and some of its features work only with its built-in models.
If something fails
| Symptom | Cause |
|---|---|
401 | The key is wrong, disabled or expired. |
402 | The payer's balance is low or the key budget is reached. Agents reserve a lot per request: set a smaller output limit in the agent. |
400 model … is not available | The model id is not in the catalog: copy it from the catalog. |
400 no route of … | The model does not support tools. Pick a model with Tools. |
| The answer stops in the middle | The output limit: set a larger max_tokens (output limit) in the agent. |