Skip to content
KoboiLLM Docs

OpenCode Integration

OpenCode is an open-source terminal-based (TUI/CLI) AI coding assistant that supports arbitrary OpenAI-compatible custom providers.

This guide shows you how to connect KoboiLLM to OpenCode for terminal-driven development, debugging, and automated refactoring using flagship models such as GPT-4o, Claude 3.5 Sonnet, or Gemini 2.5 Flash.

Step 1: Obtain API Key & Base URL

  1. API Key: Create a new API key under Virtual Keys in your KoboiLLM dashboard (sk-xxxx). Ensure All Team Models is selected.
  2. Base URL: Point to KoboiLLM’s standard OpenAI-compatible gateway: 
    https://api.koboillm.com/v1

Step 2: Configure opencode.json

OpenCode manages custom provider definitions via a JSON configuration file. You can configure this globally or per-project.

Configuration File Locations

  • Global (All Projects):
    • Linux / macOS: ~/.config/opencode/opencode.json
    • Windows: %USERPROFILE%\.config\opencode\opencode.json
  • Per-Project:
    • Create opencode.json in your project root directory.

Example Configuration

Create or edit your opencode.json file to include the KoboiLLM provider setup:

{
  "$schema": "https://opencode.ai/config.json",
  "provider": {
    "koboillm": {
      "npm": "@ai-sdk/openai-compatible",
      "name": "KoboiLLM",
      "options": {
        "baseURL": "https://api.koboillm.com/v1",
        "apiKey": "{env:KOBOILLM_API_KEY}"
      },
      "models": {
        "anthropic/claude-3-5-sonnet": {
          "name": "Claude 3.5 Sonnet",
          "limit": {
            "context": 200000,
            "output": 8192
          }
        },
        "openai/gpt-4o": {
          "name": "GPT-4o",
          "limit": {
            "context": 128000,
            "output": 16384
          }
        }
      }
    }
  }
}

Step 3: Export API Key & Run OpenCode

Export your KoboiLLM API key in your terminal session:

Terminal window
export KOBOILLM_API_KEY="sk-xxxx"
opencode --provider koboillm