Infer × Kilo Code · setup guide for 7.8.17
Add Infer to Kilo Code.
Kilo Code is built on OpenCode, and its extension and CLI read the same provider settings. Add Infer as a custom OpenAI Compatible provider in the global kilo.jsonc.
Infer Setup 0.2.0
One command.
Pick a model from the live list and enter your Infer API key when asked; the Kilo CLI starts on Infer for that session without changing your files.
npx --yes https://infer.flow7.org/downloads/infer-setup-0.2.0.tgz run kiloNeeds Node.js 20 or newer. Other clients and options.
Configuration
Choose the model and route.
The block names the route's limits and, for a route that accepts images, its image input: Kilo treats a custom model without modalities as text-only.
Waiting for the live Infer catalog…Use the global file. Kilo resolves {env:INFER_API_KEY} only in trusted config, and ignores it in a project kilo.jsonc. On Windows the file is %USERPROFILE%\.config\kilo\kilo.jsonc.
Good to know
Settings that matter.
- 01
In the extension.
Settings, Providers, Custom provider: Provider ID
infer, Provider API OpenAI Compatible, base URLhttps://infer.flow7.org/v1, and the selector as the model ID. - 02
Give the route's limits.
Without them Kilo turns compaction off and caps output at 32,000 tokens.
- 03
Tool calls on.
tool_call: truelets the agent use its tools with this model. - 04
The model is provider/model.
Kilo splits
infer/selectedat the first slash, so the selector itself goes inid.
Start with a capped key.
Create an account, add credit, and give the API key a spend limit before the first session.