Kilo Code

Kilo Code is a Cline fork with independent maintenance. It ships as a VS Code extension and a CLI; both honor an openai API provider block that points at any OpenAI-compat base URL.

Quick start

Needs: The Kilo CLI (npm install -g @kilocode/cli) and a running server (rapid-mlx serve qwen3-coder-30b-4bit).

$ rapid-mlx agents kilo-code --setup
$ kilo run "say hello"               # a reply = it works

A reply means Kilo is using your local model. rapid-mlx agents kilo-code --test runs the full integration check.

Install

$ npm install -g @kilocode/cli

Config

One-shot, with a server running: rapid-mlx agents kilo-code --setup writes the Kilo CLI's ~/.config/kilo/kilo.json, filling in the model and context window the server reports. An existing file is merged: your other keys are kept, no backup is made, and --dry-run previews without writing. The template:

// ~/.config/kilo/kilo.json
{
  "$schema": "https://app.kilo.ai/config.json",
  "model": "rapid-mlx/qwen3-coder-30b-4bit",
  "provider": {
    "rapid-mlx": {
      "npm": "@ai-sdk/openai-compatible",
      "name": "Rapid-MLX",
      "options": {
        "apiKey": "not-needed",
        "baseURL": "http://localhost:8000/v1"
      },
      "models": {
        "qwen3-coder-30b-4bit": {
          "name": "qwen3-coder-30b-4bit (Rapid-MLX)",
          "tool_call": true,
          "reasoning": true,
          "limit": {"context": 262144, "output": 8192}
        }
      }
    }
  }
}

For the VS Code extension: Settings → Kilo Code → API Provider = openai; Base URL = http://localhost:8000/v1.

Run

$ rapid-mlx serve qwen3-coder-30b-4bit
$ kilo

Recommended aliases (per rapid-mlx agents kilo-code): qwen3-coder-30b-4bit, qwen3.6-35b-4bit, qwen3.5-9b-4bit, gpt-oss-20b-mxfp4-q8.

Gotchas

• Do not point Kilo at a Cline config. Kilo is a Cline fork; the schema drifted after the split.
• First-run key prompt. Choose "OpenAI Compatible" and paste not-needed as the API key.
• Continue.dev migrants: Kilo's JSON config is stricter. Quote all API base URLs.

Empirical

The kilo-code row of the integration matrix is ✅ on Qwen 3.6, Gemma 4, and gpt-oss; DeepSeek R1-Distill XFAIL(arch). Source of truth for this page: TestKiloCode + rapid-mlx agents kilo-code.

See also