GitHub Pages

oh-my-codex

Multi-agent orchestration for OpenAI Codex CLI.

npm install -g oh-my-codex

What's New in 0.6.0

Mixed Codex + Claude teammates

Route workers per pane with OMX_TEAM_WORKER_CLI_MAP (for example: codex,codex,claude,claude) so one $team run can use both CLIs.

Claude-team leader nudges

Added a leader-side all-workers-idle fallback so notifications still fire even when worker-side Codex hooks are unavailable.

Safer trigger retries

Team trigger fallback now gates adaptive resend behind a ready prompt and no-active-task check, with safer clear-line resend behavior.

Reliability fixes

Improved task-claim bootstrap, stricter CLI map validation, predictable auto resolution, and better leader pane targeting for mixed/Claude teams.

Core Features

Agent Prompts

Specialized roles for implementation, review, design, and planning.

Skills

Composable workflows from autopilot and team pipelines to setup, release, and tracing.

Team Orchestration

Staged execution pipeline with verification and fix loops.

MCP State Management

Persistent lifecycle state, notepad, and project memory across sessions.

Quick Start

  1. npm install -g oh-my-codex
  2. omx setup
  3. omx doctor
  4. omx --xhigh --madmax (trusted environments) or omx

Integration Guides

OpenClaw / notification gateway setup and prompt-tuning guides by language:

Open dedicated Integrations page

Architecture

User
  -> Codex CLI
     -> AGENTS.md (orchestration brain)
     -> ~/.codex/prompts/*.md (installable active/internal agent prompt catalog)
     -> ~/.codex/skills/*/SKILL.md (skill catalog)
     -> ~/.codex/config.toml (features, notify, MCP)
     -> .omx/ (runtime state, memory, plans, logs)