██████╗ ██╗ ██╗ ███╗ ███╗██╗ ██╗
██╔═══██╗██║ ██║ ████╗ ████║╚██╗ ██╔╝
██║ ██║███████║ ██╔████╔██║ ╚████╔╝
██║ ██║██╔══██║ ██║╚██╔╝██║ ╚██╔╝
╚██████╔╝██║ ██║ ██║ ╚═╝ ██║ ██║
╚═════╝ ╚═╝ ╚═╝ ╚═╝ ╚═╝ ╚═╝
██████╗ ██████╗ ██████╗ ██╗██╗ ██████╗ ████████╗
██╔════╝██╔═══██╗██╔══██╗██║██║ ██╔═══██╗╚══██╔══╝
██║ ██║ ██║██████╔╝██║██║ ██║ ██║ ██║
██║ ██║ ██║██╔═══╝ ██║██║ ██║ ██║ ██║
╚██████╗╚██████╔╝██║ ██║███████╗╚██████╔╝ ██║
╚═════╝ ╚═════╝ ╚═╝ ╚═╝╚══════╝ ╚═════╝ ╚═╝
Turbocharge your Copilot CLI with multi-agent orchestration
Your Copilot has been working out, learning new ways to improve your life.
This work is based on oh-my-claudecode by Yeachan Heo, but with copilot cli focus.
# Step 1: Install
/plugin marketplace add https://github.com/RobinNorberg/oh-my-copilot
/plugin install oh-my-copilot@omcp
# or
npm i -g oh-my-copilot@latest
# Step 2: Setup
/setup
# Step 3: Build something
/autopilot: build a todo-app
# or
autopilot: build a todo-app
# If you enjoy the output, give the repo att ⭐ and tell a friendIf you're uncertain about requirements, have a vague idea, or want to micromanage the design:
/deep-interview "I want to build a todo-app"
The deep interview uses Socratic questioning to clarify your thinking before any code is written.
- Copilot CLI friendly — Perfect for development in the terminal
- Natural language interface — no commands to memorize, just describe what you want
- Team-first orchestration — staged pipeline with plan, exec, verify, and fix loop
- Automatic parallelization — complex tasks distributed across our specialized agents
- Smart model routing — Haiku for simple tasks, Sonnet for average and Opus for complex reasoning
- Persistent execution — won't give up until the job is verified complete
- Azure DevOps/GitHub native — auto-detection, work item management, PR operations, triage workflows
- Stop your yolo abuse — using a layered permission model to help your agents perform safe work without your interference
Optional shortcuts for power users. Natural language works fine without them.
| Keyword | Category | Effect | Example |
|---|---|---|---|
team |
 |
Canonical Team orchestration | team 3:executor "fix all TypeScript errors" |
ask copilot |
|
Delegate to Copilot CLI | ask claude "review auth architecture" |
ask claude |
|
Delegate to Claude Code CLI | ask claude "review auth architecture" |
ask codex |
|
Delegate to Codex CLI | ask codex "security analysis" |
ask gemini |
|
Delegate to Gemini CLI | ask gemini "suggest UX improvements" |
cccg |
|
Quadri-model orchestration | cccg review this PR |
omcp team |
|
tmux CLI workers (codex/gemini/copilot) | omcp team 2:codex "security review" |
code review |
 |
Code review mode | code review the auth module |
critique |
|
Pre-push adversarial critique | critique my changes |
debug, diagnose |
|
Session/repo diagnostics | debug why hooks aren't firing |
deep-analyze |
|
Deep analysis mode | deep-analyze why tests are failing |
deep-dive |
|
Trace → interview pipeline | deep-dive why auth is slow |
deepinit |
|
Deep codebase init with AGENTS.md | deepinit |
deep-interview |
|
Socratic requirements clarification | deep-interview "vague idea" |
deep-review |
|
Multi-pass code review (4 passes) | deep-review this PR |
deepsearch |
|
Codebase-focused search routing | deepsearch for auth middleware |
deslop, anti-slop |
|
AI code slop cleanup | deslop the auth module |
discover |
|
Parallel codebase quality scan | discover src/hooks/ |
external-context |
|
Parallel external doc/web search | external-context React Server Components |
sciomc |
|
Parallel scientist orchestration | sciomc analyze test failures |
security review |
|
Security review mode | security review the API endpoints |
tdd, test first |
|
TDD workflow enforcement | tdd: implement password validation |
trace |
|
Evidence-driven causal tracing | trace why auth is broken |
verify |
|
Verify changes work before claiming done | verify the fix |
ultrathink |
|
Deep reasoning mode | ultrathink about this architecture |
ralplan |
 |
Iterative planning consensus | ralplan this feature |
autopilot |
 |
Full autonomous execution | autopilot: build a todo app |
cancelomc, stopomc |
|
Stop active OMC modes | stopomc |
experiment |
|
Hypothesis-driven experiment loop | experiment: optimize API latency |
ralph |
|
Persistence mode | ralph: refactor auth |
ralphthon |
|
Autonomous hackathon mode | ralphthon: build MVP in 2 hours |
self-improve |
|
Autonomous evolutionary code improvement | self-improve the parser module |
skillify |
|
Extract reusable skill from session | skillify this workflow |
ulw |
|
Maximum parallelism | ulw fix all errors |
gh setup |
 |
Configure GitHub integration | gh setup |
gh triage |
|
GitHub issue/PR/CI triage | gh triage |
gh review |
|
Interactive GitHub PR review | gh review |
gh auto-review |
|
Automated code review via code-reviewer agent | gh auto-review |
gh project |
|
Manage GitHub Projects (v2) boards | gh project |
ado sprint |
 |
Sprint planning and iteration management | ado sprint |
ado setup |
|
Configure Azure DevOps integration | ado setup |
ado triage |
|
Azure DevOps work item triage | ado triage |
ado review |
|
Interactive Azure DevOps PR review | ado review |
ado auto-review |
|
Automated code review via code-reviewer agent | ado auto-review |
hud |
 |
Configure status line display | hud preset minimal |
learner |
|
Learn a skill from the current conversation | learner |
remember |
|
Save reusable project knowledge | remember this pattern |
wiki |
|
Persistent markdown knowledge base | wiki add auth architecture notes |
Notes:
- Informational filtering: Asking "what is ralph?" or "explain ultrawork" won't trigger execution — only actionable uses activate keywords.
Team is the canonical orchestration surface. It runs a staged pipeline:
team-plan → team-prd → team-exec → team-verify → team-fix (loop)
/team 3:executor "fix all TypeScript errors"CCCG uses multi-model advisor synthesis — fans out via ask-copilot + ask-claude + ask-codex + ask-gemini, then Copilot synthesizes the results:
/cccg "review this branch — architecture, security, and UI components"Omc team spawn real tmux CLI workers for cross-model tasks:
omcp team 1:copilot "review the ingestion module for performance issues"
omcp team 2:claude "review the database module for sql issues"
omcp team 3:codex "review the auth module for security issues"
omcp team 5:gemini "redesign UI components for accessibility"- Documentation Home
- Quick Start
- Full Reference
- Team Mode
- Azure DevOps Integration
- GitHub Integration
- Architecture Overview
- Permission Architecture
OMC features like omcp team and rate-limit detection require tmux:
| Platform | tmux provider | Install |
|---|---|---|
| macOS | tmux | brew install tmux |
| Ubuntu/Debian | tmux | sudo apt install tmux |
| Fedora | tmux | sudo dnf install tmux |
| Arch | tmux | sudo pacman -S tmux |
| Windows | psmux (native) | winget install psmux |
| Windows (WSL2) | tmux (inside WSL) | sudo apt install tmux |
Windows users: psmux provides a native
tmuxbinary for Windows with 76 tmux-compatible commands. No WSL required.
OMC can orchestrate multiple AI CLI providers as tmux workers for cross-validation, design consistency, and parallel execution. All four major CLI tools are supported:
| Provider | Install | What it enables |
|---|---|---|
| Copilot CLI | npm install -g @github/copilot |
Core orchestration platform |
| Claude Code | npm install -g @anthropic-ai/claude-code |
Deep reasoning, architecture analysis |
| Gemini CLI | npm install -g @google/gemini-cli |
Design review, UI consistency (1M token context) |
| Codex CLI | npm install -g @openai/codex |
Architecture validation, code review cross-check |
omcp team 2:claude "review auth architecture"
omcp team 2:codex "security analysis"
omcp team 2:gemini "UI consistency check"
omcp team 1:copilot "review existing tests"Only Copilot CLI is required — the others are optional and add cross-provider validation.
Inspired by: • oh-my-claudecode, oh-my-opencode • Superpowers • get-shit-done • Ouroboros • BMAD
Contributions welcome! Open an issue or PR on GitHub to the dev branch.
MIT — see LICENSE.