Skip to content

OpenCode Agent

The OpenCode Agent is an open-source, terminal-first AI coding agent. It allows developers to interact with artificial intelligence directly from the command line interface (CLI) to generate code, troubleshoot issues, analyze repositories, or automate terminal tasks.

As a model-agnostic, open alternative to proprietary tools (like Claude Code), OpenCode integrates seamlessly with our Swiss-hosted LLMaaS Gateway. This ensures that all of your code context, prompts, and sensitive data remain entirely within our highly secure Swiss infrastructure.

Key Features

  • Terminal-First Workflow: Interact with an interactive Terminal User Interface (TUI) directly within your workspace.
  • Full Data Sovereignty: By routing traffic through our Swiss LLMaaS Gateway, your prompts and source code never leave Switzerland.
  • Model-Agnostic: Flexibly switch between different models and providers.
  • Agentic Capabilities: The agent can independently analyze project structures, edit files, execute tests, and fix programming bugs interactively.
  • Context Branching: Allows you to "rewind" a conversation to explore alternative implementation paths without losing the original conversation history.

Installation

OpenCode can be installed easily using an installation script or package manager on macOS and Linux:

curl -fsSL https://opencode.ai/install | bash

brew install opencode

npm install -g @opencode/cli

Configuration with LLMaaS Gateway

To connect OpenCode to our Swiss LLMaaS endpoint, you can create a configuration file at the project level or globally in your user directory.

1. Save API Key

Export your virtual API key as an environment variable. Replace sk-bf-... with your actual key:

mkdir -p ~/.config/opencode
echo "sk-bf-..." > ~/.config/opencode/ew-key

2. Create Configuration File

Create a file named opencode.json in your project directory (or globally at ~/.config/opencode/opencode.json):

{
  "$schema": "https://opencode.ai/config.json",
  "enabled_providers": ["ew"],
  "provider": {
    "ew": {
      "npm": "@ai-sdk/openai-compatible",
      "name": "ai.ewcs.ch",
      "options": {
        "baseURL": "https://ai.ewcs.ch/v1",
        "apiKey": "{file:~/.config/opencode/ew-key}"
      },
      "models": {
        "ew/minimax27": {
          "name": "minimax27"
        },
        "ew/qwen36-35b-a3b": {
          "name": "qwen36-35b-a3b"
        },
        "ew/inference-deepseek-v32": {
          "name": "deepseek-v32"
        },
        "ew/inference-deepseekr1-670b": {
          "name": "deepseekr1-670b"
        },
        "ew/inference-gemma4-31b": {
          "name": "gemma4-31b"
        },
        "ew/inference-gpt-oss-120b": {
          "name": "gpt-oss-120b"
        },
        "ew/inference-kimi-k2": {
          "name": "kimi-k2"
        },
        "ew/inference-llama4-maverick": {
          "name": "llama4-maverick"
        }
      }
    }
  }
}

3. Connect

Start the agent in your project directory:

opencode

Within the interactive terminal, run the /connect command, select the configured everyware-llmaas profile, and select the desired model using the /models command.

Typical Commands and Workflows

Once OpenCode is running, you can enter prompts or slash commands directly in the terminal interface:

  • Code Analysis: "Explain the architecture of this repository and show me where the API calls are made."
  • Refactoring: "Optimize error handling in script/translate_md.py and add unit tests."
  • Run Commands: "Run the linter and fix all reported syntax errors."
  • TUI Commands:
    • /connect – Manage connections to providers
    • /models – Switch the active LLM model
    • /clear – Clear chat history
    • /exit – Exit the OpenCode TUI

Extension: Oh My OpenAgent (OmO)

Learn how to extend OpenCode with Oh My OpenAgent (OmO) to support multi-agent workflows (agent teams) and intelligent model routing.