Connect Kilo Code to Nbility
Kilo Code's OpenAI Compatible provider can connect to Nbility. This guide covers the editor extension; consult the matching official documentation for the standalone CLI.
Install
Install Kilo Code from the VS Code marketplace, or run:
code --install-extension kilocode.kilo-code
For VSCodium and other compatible editors, use the extension source recommended by Kilo.
Configure the connection
Open the Kilo Code sidebar settings and create or edit a provider profile:
| Field | Value |
|---|---|
| API Provider | OpenAI Compatible |
| Base URL | https://api.nbility.ai/v1 |
| API key | A dedicated Nbility token |
| Model ID | An exact identifier from Models |
Prefer the Base URL field. Use https://api.nbility.ai/v1/chat/completions only when your version explicitly requests a complete endpoint. Do not mix these fields.
Check capabilities and test
Set tools, images, context size, and output limits according to the actual model. Start with a tool-capable model, save the profile, and confirm it is active.
Create a task asking the model to read and explain a small file. After the response and tool calls work, confirm the request in Usage logs, then try a small edit.
Troubleshoot
- Missing model: enter the complete ID manually and check the token group.
- 404: look for a complete endpoint placed in the Base URL field, duplicating the path.
- Tool error: check capability options and select a model supporting the required call format.
- Old model after changing profile: confirm the profile active for the task or start a new conversation.
See Connection basics for general troubleshooting.
References
Informed by the aicentos Kilo Code tutorial (CC BY 4.0), rewritten for Nbility. See Kilo's OpenAI Compatible documentation for current options.