Introduction
With the advancement of AI-assisted development environments, developer productivity has dramatically improved. Among the growing demand for development support tools leveraging GitHub Copilot, opencode and oh-my-opencode are particularly noteworthy tools.
This article explains how to install opencode and its extension tool oh-my-opencode in a WSL environment, and utilize various AI models such as Claude Sonnet 4.5 and GPT-5.2 as agents. This enables you to use multiple AI models according to different purposes and maximize development efficiency.
What is opencode?
opencode is a command-line tool that allows you to directly use AI assistants from the terminal. It integrates with various AI providers including GitHub Copilot, supporting tasks such as code generation, refactoring, and documentation.
Features of oh-my-opencode
oh-my-opencode is a framework that extends opencode and enables management and customization of multiple AI agents. It can automatically select and execute the optimal AI model according to the project nature and work content.
Key features include:
- Multi-agent management: Configure multiple AI models by purpose
- Automated task execution: Automate from planning to code generation
- Flexible permission settings: Fine-grained control over file editing and command execution permissions
Installation Steps
Prerequisites
This procedure assumes the following environment:
- WSL (Windows Subsystem for Linux) is installed
- GitHub Copilot subscription is active
- Internet connection is available
1. Installing opencode
Open a WSL terminal and execute the following command:
curl -fsSL https://opencode.ai/install | bash
This script installs the necessary dependencies and adds the opencode command to your system. After installation is complete, start a new shell session or reload the path with the following command:
source ~/.bashrc
Next, launch opencode:
opencode
After launch, when the prompt appears, enter the /connect command and select GitHub Copilot as the provider. Follow the on-screen instructions to complete authentication, establishing the connection with GitHub Copilot.
2. Installing oh-my-opencode
With opencode running, enter the following command:
Install and configure by following the instructions here https://raw.githubusercontent.com/code-yeongyu/oh-my-opencode/refs/heads/master/README.md
This command executes the automated setup described in the oh-my-opencode README. Follow the prompts to complete the necessary configuration.
3. Configuring Individual AI Models
In oh-my-opencode, you can specify the AI model to use for each agent. This allows you to leverage the optimal model according to the task.
The configuration file path is as follows:
~/.config/opencode/oh-my-opencode.json
Open this file in a text editor and edit it based on the following content:
{
"$schema": "https://raw.githubusercontent.com/code-yeongyu/oh-my-opencode/master/assets/oh-my-opencode.schema.json",
"experimental": {
"auto_resume": true
},
"google_auth": false,
"sisyphus_agent": {
"default_builder_enabled": true,
"planner_enabled": true
},
"agents": {
"Sisyphus": {
"model": "github-copilot/claude-opus-4.5"
},
"oracle": {
"model": "github-copilot/gpt-5.2"
},
"librarian": {
"model": "github-copilot/grok-code-fast-1"
},
"explore": {
"model": "github-copilot/gemini-3-flash-preview"
},
"frontend-ui-ux-engineer": {
"model": "github-copilot/gemini-3-pro-preview"
},
"document-writer": {
"model": "github-copilot/gemini-3-pro-preview"
},
"multimodal-looker": {
"model": "github-copilot/gemini-3-flash-preview"
},
"*": {
"permission": {
"edit": "allow",
"webfetch": "allow",
"doom_loop": "allow",
"external_directory": "allow",
"bash": {
"*": "allow",
"git": "allow",
"rm": "ask"
}
}
}
}
}
Here's an explanation of each agent's configuration:
- Sisyphus: Handles complex development tasks and planning. Uses Claude Opus 4.5
- oracle: Handles advanced reasoning and problem-solving. Uses GPT-5.2
- librarian: Handles code search and documentation reference. Uses Grok Code Fast 1
- explore: Handles exploratory research and verification. Uses Gemini 3 Flash Preview
- frontend-ui-ux-engineer: Handles frontend development and UI/UX design
- document-writer: Handles documentation creation and article writing
In the permission settings (permission), rules are defined for when each agent performs file editing, web fetching, and external command execution. Setting only the rm command to ask helps prevent accidental deletions.
4. Installing Dependencies
oh-my-opencode agents use tmux to manage terminal sessions. Install it with the following command:
sudo apt install tmux
By installing tmux, agents can execute multiple terminal sessions in parallel and process tasks efficiently.
Verification
After installation is complete, verify the operation with the following steps:
- Launch
opencode:
opencode
-
At the prompt, enter a command to invoke an agent (e.g.,
@Sisyphus Execute the task). -
Verify that the agent responds normally and the configured model (e.g.,
github-copilot/claude-opus-4.5) is being used. -
Check the logs to ensure no errors have occurred.
Troubleshooting
When Model is Not Found
Verify that the model identifier specified in the configuration file is correct. Since models available via GitHub Copilot are updated regularly, you need to check for the latest information.
Access Models.dev, search for github-copilot, and obtain the latest model identifiers. Then update the oh-my-opencode.json file and restart opencode.
When Authentication Error Occurs
GitHub Copilot authentication may have failed. Try re-authenticating with the following steps:
- Launch
opencodeand disconnect the provider with the/disconnectcommand. - Execute the
/connectcommand again, select GitHub Copilot, and complete authentication.
When tmux is Not Found
If tmux is not installed, agents will not work properly. Complete the installation with the command mentioned earlier.
Summary
This article explained the installation steps for opencode and oh-my-opencode in a WSL environment. By utilizing these tools, you can use multiple AI models via GitHub Copilot and significantly improve development efficiency.
By assigning appropriate models to each agent, you can receive optimal support according to the task. Please try using them in your actual projects.
Related Links
Related Articles
Building a Self-Improving Advisor System with GitHub Copilot and Knowledge Base
Built a self-improving advisor system using GitHub Copilot's agent functionality that evaluates and improves a knowledge base. Multiple agents collaborate to automatically enhance response quality.
How to Configure Copilot Automated Code Review with GitHub Rulesets
This article explains how to use GitHub Rulesets to enable automated code reviews by GitHub Copilot for pull requests targeting specific branches.
Codex CLI: A Complete Guide to the Latest Settings and Features (GPT-5 Ready)
With GPT-5 support, the Codex CLI is rapidly gaining attention. This guide covers everything from basic setup to handy tips and important notes for Windows users.