Oh My OpenAgent (OmO) 🤖 Agent Free

What it does

Oh My OpenAgent (OmO), previously called oh-my-opencode, is a multi-agent harness plugin. It turns a single coding-agent session into an orchestrated team that works across several models. You type ultrawork (or ulw) with your prompt. A main orchestrator then plans the work, hands pieces to specialist subagents chosen by category, such as visual work, deep work, quick edits or hard logic, and keeps going until the task list is finished. It assigns models from different providers to each role, so it is not tied to one vendor's subscription.

What is inside

• Discipline agents: the orchestrator, an interview-style planner (/ulw-plan) and background specialists that run in parallel
• Team Mode (opt-in): a lead agent with up to eight parallel members shown in tmux, which also powers the hyperplan critique and security-research skills
• Tooling: LSP rename, definition and diagnostics, AST-Grep search and rewrite, tmux terminals, and optional hash-anchored line edits
• Built-in MCP servers: web search (Exa), Context7 docs, grep.app code search and LSP
• Guards and helpers: a todo enforcer, a comment checker that removes low-value AI comments, AGENTS.md rules injection and /init-deep for hierarchical AGENTS.md files
• Claude Code compatibility: existing hooks, commands, skills and MCP servers load unchanged

Works with

The Ultimate edition is a plugin for OpenCode. A Light edition brings a subset to OpenAI Codex CLI. A standalone omo beta edition runs on the Pi agent engine.

How to install

bunx oh-my-openagent install

For Codex, run npx lazycodex-ai install instead. The README also suggests asking an agent to follow the installation guide.

Maintenance and safety

The project is very actively developed, but it is released under the Sustainable Use License, a source-available licence that is not OSI-approved. Anonymous telemetry through PostHog is on by default and can be turned off with a config flag or environment variable. The installer edits your OpenCode or Codex config, asks for provider credentials, and can set Codex to autonomous full-permission mode, so read each prompt before accepting. Long unattended loops can use a lot of tokens.

Who should use it

Experienced OpenCode or Codex users who want aggressive, parallel multi-model orchestration and are comfortable tuning a large, fast-moving configuration.