engineering / skill

oh-my-claudecode

Adds multi-agent orchestration, such as autopilot and team modes, to Claude Code through slash commands and a terminal CLI.

Repository owner Yeachan-Heo

Not rated yet
0 comments

Like 0 Save

Published 11 Oct 2026Listing updated 12 Oct 2026

At a glance

Purpose: Adds multi-agent orchestration, such as autopilot and team modes, to Claude Code through slash commands and a terminal CLI.
Runs in: Local agent with shell, MCP server
Operating systems: macOS, Linux, Windows
Needs: Runtime: tmux for team and rate-limit features; Needs internet: Not stated; Needs accounts or API keys beyond your agent: Not stated
Files and data: Reads files: Yes; Writes files: Yes; Sends data to other services: Yes
Tool type: Skill
Delivery: From GitHub
Version: Latest on the main branch
Access: Code on GitHub
Compatibility: According to the README, it is built for Claude Code; its team-worker CLI additionally supports Codex, Gemini CLI, Antigravity, Grok Build CLI, and Cursor, and it exports Node.js SDK helpers for @anthropic-ai/claude-agent-sdk programs.
External costs: The repository is MIT-licensed and states no price for itself. It runs through your own Claude Code and Anthropic usage, and, according to the README, using its CLI workers for Codex, Gemini, Antigravity, Grok, or Cursor needs that provider's own CLI and its own subscription or API key.
Permissions: The skills act through your coding agent with the access you give it. According to the README, its terminal CLI can also run shell commands, including tmux panes and other providers' CLIs, and it writes runtime state, session data, plans, logs and handoffs to a local .omc/ directory (or ~/.omc/ outside a git repository), which it says may contain prompts and transcripts. It says to enable native Team mode by editing ~/.claude/settings.json, and omc wait --start enables an auto-resume daemon for Claude Code sessions after rate limits reset.
Usage terms: MIT License, as stated in the repository. See its LICENSE file for the exact terms.

Code on GitHub

Yeachan-Heo/oh-my-claudecode

MIT

TypeScript39,839 stars3,567 forksLast pushed 13 Oct 2026 UTC

Teams-first Multi-agent orchestration for Claude Code

Topics: agentic-coding ai-agents automation claude claude-code multi-agent-systems oh-my-opencode opencode

Public GitHub data fetched 13 Oct 2026 UTC · 0 days old

View on GitHub ↗

What it does

oh-my-claudecode, abbreviated OMC, is a multi-agent orchestration add-on for Claude Code, installed as a plugin or an npm-based terminal CLI. According to the README, its canonical surface is Team, a staged pipeline of plan, PRD, execution, verification, and fix stages, alongside other modes such as Autopilot for autonomous single-lead execution, Execute for a persistent verify-and-fix loop, Verify for repeated quality-gate checks, and Deep Interview for Socratic requirements clarification.

The repository bundles 47 skills, covering orchestration (autopilot, execute, team, verify, ralph, ralplan), planning and research (plan, research, autoresearch, architecture-survey), and project utilities (debug, review, diagram, wiki, skillify, remember). According to the README, its terminal CLI can also launch tmux panes that run other providers' own CLIs, including Codex, Gemini, Antigravity, Grok, or Cursor, as team workers alongside Claude.

According to the README, you install the plugin, run its setup command, then trigger a mode with a slash command inside a Claude Code session, such as /autopilot "build a REST API for managing tasks". A separate omc terminal command covers tasks such as setup, provider questions and tmux team workers; the README says Autopilot, Ralph, Execute and Deep Interview run only inside a session. It also includes a HUD status line, cost tracking, and a /skillify command that extracts reusable skills from past sessions.

Who it is for

Developers using Claude Code who want multi-agent orchestration and automated feature workflows.

Example tasks

Run /autopilot with a feature description for autonomous end-to-end execution.
Start a /team session with several executor agents to fix a batch of errors.
Ask /deep-interview to clarify a vague feature idea before any code is written.
Launch a cross-model review with omc team 2:codex "review auth module for security issues".
Extract a reusable skill from a past debugging session with /skillify.

Getting started

The README's plugin route is "/plugin marketplace add https://github.com/Yeachan-Heo/oh-my-claudecode", then "/plugin install oh-my-claudecode", then "/omc-setup". Its CLI route is "npm i -g oh-my-claude-sisyphus@latest", then "omc setup". The README has the exact commands.

Limitations

Needs Claude Code itself. The terminal CLI path installs through npm with "npm i -g oh-my-claude-sisyphus@latest". Per the README, native Team mode needs the CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS setting enabled, CLI team workers for Codex, Gemini, Antigravity, Grok, or Cursor each need that provider's own CLI installed and authenticated plus an active tmux session, and named autopilot stage profiles need Linux with the flock utility. The repository can change at any time; check its README before you install.

Data practices

The README gives no single data-handling statement. It says runtime state, session data, plans, logs and handoffs are written locally under .omc/, which it says may contain prompts, transcripts or machine-specific paths. It also says stop callbacks can send session summaries to Telegram, Discord or Slack, and an OpenClaw integration can forward Claude Code session events to a gateway, when you configure them. It describes an optional Jev feature that queries an API with a key you set, and says that without configuration nothing is sent to it. The README does not describe what is sent to Claude, Codex, Gemini, Antigravity, Grok or Cursor; that depends on which of them you use.