AI
Omarchy ships with OpenCode and Claude Code as the default agent harnesses for programmers. OpenCode lets you use models from all the major commercial providers as well as the rapidly improving open-weight models.
The best way to use OpenCode is by going to the directory you're going to work on in the terminal, then invoke the c alias to start OpenCode within that directory. Claude Code let's you use Anthropic's models with their proprietary subscriptions, and can be started in danger mode using the cx alias.
Alternative agent CLIs
Beyond the default c (OpenCode) and cx (Claude Code) aliases, Omarchy ships pre-wired lazy-loaded launchers for every major coding-agent CLI. The launchers themselves are tiny shell stubs in ~/.local/bin/, so on first run, each one fetches its npm package through npx against a mise-managed node@latest, so nothing is downloaded until you actually invoke it. Run any of these and authenticate when prompted:
| Command | Package | Provider |
|---|---|---|
codex |
@openai/codex | OpenAI Codex |
gemini |
@google/gemini-cli | Google Gemini CLI |
copilot |
@github/copilot | GitHub Copilot CLI |
pi |
@mariozechner/pi-coding-agent | Mario Zechner's Pi |
To wrap an additional npm-published CLI the same way, run omarchy-npx-install <package> [command-name].
Local LLMs
Omarchy recommends two ways of running local LLM models: LM Studio and Ollama. LM Studio provides a GUI interface for finding open-weight models, installing them, and running them. It's a great way to get going easily. Ollama offers a CLI for doing so similarly. But if you're new to local models, I'd start with LM Studio. You can install either under Install > AI in the Omarchy Menu.
The Omarchy Skill
Agent skills help AI use specific tools in a specific way, and Omarchy ships with a default skill for tailoring the system. Like making changes to waybar or even creating a new theme from scratch. It's automatically configured in OpenCode (and Claude Code and any other harness that supports ~/.claude/skills).
But you should treat this skill as experimental. Different models will use it to different effect. It's best to run in plan mode first, so you have an idea of what the agent would like to change. And then be ready to rollback changes or even invoking omarchy reinstall configs, if the agent makes a mess of everything.