oh-my-codex (OMX)
github.com/yeachan-heo/oh-my-codex- Category
- AI Agents
- Rank
- No. 180Tools index
- Pricing
- Open Source
- Type
- TOOL
- Builder
- yeachan-heo
- GitHub
- 33.0k stars
- Latest release
- v0.21.4
- Date
About
A workflow layer for OpenAI Codex CLI that adds better prompts, agent teams, runtime help, and project state management. It enhances Codex with canonical skills like deep-interview, ralplan, team coordination, and persistent completion loops.
What it does
OMX sits in front of an authenticated Codex command and organizes longer coding jobs into explicit clarification, planning, execution, review, and verification stages. It can launch isolated worktrees, coordinate parallel workers through terminal panes, display live state, and preserve plans, logs, memory, and workflow progress inside the project.
Why it's ranked here
The appeal is operational depth, not a replacement coding engine. OMX provides scoped setup, health checks, real authentication smoke tests, resumable pipeline state, isolated worktree launches, and guarded concurrent sessions. Those controls make it credible for long or risky Codex jobs, although the recommended setup assumes substantial commitment to its conventions and terminal environment.
What's good
The project treats orchestration as durable infrastructure. Planning and execution produce persistent artifacts, ordinary sessions fail closed when another launch owns the writable pointer, and dangerous direct launches require approval. Health checks distinguish installation problems from failures that appear only during a real model request. The package also exposes diagnostics, cleanup, session search, authentication slots, a live status display, and team monitoring.
Tradeoffs
OMX requires Node.js 20 or newer plus a working, authenticated Codex installation. Its preferred team experience also depends on tmux. Native Windows and the Codex App receive less support and may behave inconsistently. Setup has project and user scopes, persistent guidance, plugin interactions, merge policies, update behavior, and many workflow modes, so adopting it adds concepts and state that plain Codex users may not want.
How to use it well
Use OMX for long, staged repository work where requirements need clarification, plans need approval, and completion must survive multiple steps or coordinated workers. Start in a task-specific worktree, run both the installation check and live model smoke test, then use team execution only for genuinely parallel stories. It does not provide Codex itself, authentication, or a substitute for a supported terminal environment.
Technical notes+
package.json defines an ES module npm package requiring Node.js 20+, with TypeScript compilation, Biome linting, Node tests, coverage scripts, packed-install smoke tests, plugin synchronization, and Rust build tasks. Cargo.toml declares a six-member Rust workspace using edition 2021 and Rust 1.73. src/index.ts exposes setup, diagnostics, configuration, agent definitions, HUD, and VS Code helpers as a library surface. src/cli/index.ts implements the broad command router and launch policies. src/pipeline/index.ts models the staged autopilot sequence. src/hud/index.ts and src/sidecar/index.ts implement terminal monitoring, including tmux integration. src/vscode/index.ts adds direct-session spawning, dangerous-argument approval, environment resolution, and redacted logs. src/url-reader/index.ts validates protocols, DNS results, redirects, address ranges, response sizes, and challenge markers before returning remote text.
Observed
- License
- MIT
- Primary implementation
- TypeScript ES modules with a Rust workspace
- Install surface
- Global npm package named oh-my-codex, exposing the omx command
- Runtime requirements
- Node.js 20 or newer and an installed, authenticated Codex CLI
- Interfaces
- Command-line tool, JavaScript library exports, Codex plugin layout, and MCP server support
- Platform support
- macOS and Linux are the recommended path; native Windows and Codex App receive less support
- Native workspace
- Six Rust workspace crates using edition 2021 and Rust 1.73
Read from README.md, Cargo.toml, package.json, src/index.ts, src/cli/index.ts, src/hud/index.ts, src/auth/index.ts, src/wiki/index.ts, src/adapt/index.ts, src/vscode/index.ts, src/sidecar/index.ts, src/openclaw/index.ts, src/pipeline/index.ts, src/url-reader/index.ts.
What it can do
Enhance OpenAI Codex prompts for better code generation
Natural language prompts and coding requests → Improved prompts optimized for Codex execution
Coordinate AI agent teams for complex tasks
Multi-step project requirements → Coordinated workflow execution across multiple AI agents
Conduct deep technical interviews
Interview parameters and candidate information → Structured technical interview sessions and evaluations
Generate project plans using ralplan methodology
Project requirements and constraints → Structured project plans and implementation roadmaps
Manage persistent project state across sessions
Project files and workflow data → Maintained state files in .omx/ directories
Execute completion loops for iterative development
Code generation tasks and feedback → Refined code through multiple iteration cycles
Provide runtime assistance during coding sessions
Active coding context and user queries → Contextual help and guidance during development
Tags
Tech Stack
Media
Comments (0)
No comments yet
Editorially curated, with community endorsements as a secondary signal. Corrections welcome.