codexclaw plugin
A Codex plugin adds development discipline, 29 skills, 31 hooks and multi-model subagents in one package.
- 54 ★ on the repository
- Last commit
- MIT
Extended access
Why this level
- The plugin installs 31 hooks that run in the Codex environment.
- The auto-approve setting for agent-created threads removes part of the approval prompts.
Install
- Works in
Manual install
codex plugin marketplace add https://github.com/lidge-jun/codexclawThen `codex plugin add codexclaw@codexclaw`, restart Codex and approve the 31 hooks.
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- The plugin installs 31 hooks that run in the Codex environment.
- The auto-approve setting for agent-created threads removes part of the approval prompts.
This is third-party code. Review the repository files before installing.
What it does
codexclaw turns Codex into a disciplined development environment. It has no agent harness of its own, and skills, hooks and components layer directly on `codex`. A `dev` family of 12 surface routers, including `dev-architecture`, `dev-backend`, `dev-testing`, `dev-security` and `dev-debugging`, sets rules for each area. The PABCD workflow (plan, audit, build, check, done) is a file-backed state machine with attested transitions through `cxc orchestrate`. Subagents split by role: explorer, reviewer, executor and architect, each with its own model and prompt. Plan and ledger state can be kept outside the repository in `~/.codexclaw/projects/`.
Who it is for. Codex developers who want structured work phases, attested transitions and subagent roles on different models.
Good fit when
- Before a code change you need a diff-level plan with an interview and an audit.
- An explorer, a reviewer and an executor run on different models by role.
- Plans and ledgers of an enterprise repository are kept outside the repository.
- The PABCD cycle status is checked with `orchestrate status`.
Not a fit when
- You do not work in Codex, since the plugin is built for it.
- You want minimal setup with no hooks, as the plugin installs 31 hooks that need approval.
- You need an official product, since the project is unofficial and independent of OpenAI.
Example request
Plan the payment module migration with codexclaw PABCD and multi-model subagents: interview first, then a diff-level plan.Limitations
After installation you must restart Codex and approve 31 hooks, and upgrades ask again by content hash. The project is unofficial: it builds on ideas from oh-my-openagent, and some components are adapted from MIT-licensed projects listed in `NOTICE.md`. The `agentCreatedThreadAutoAllow` setting removes approval prompts in agent-created threads and should be enabled only knowingly. State location needs separate configuration for enterprise repositories. Documentation is available in English, Korean and Chinese.
How to disable. Remove the plugin with Codex's plugin command and follow the update and uninstall section of the README. Clear `~/.codexclaw/` and the `devlog/` and `.codexclaw/` folders separately.
FAQ
How do I install it?
With two commands: `codex plugin marketplace add https://github.com/lidge-jun/codexclaw` and `codex plugin add codexclaw@codexclaw`, then a restart and hook approval.
What is PABCD?
A cycle of Plan, Audit, Build, Check and Done phases with gates between them. Transitions are attested with evidence through `cxc orchestrate`.
What roles do subagents have?
Explorer, reviewer, executor and architect. Each role can have its own model and prompt, and the configuration persists across sessions.
Where is state stored?
In the repository by default, but outside it under `~/.codexclaw/projects/` for repositories with a foreign owner. `CODEXCLAW_STATE_MODE` sets the mode.
Is the project affiliated with OpenAI?
No, the project is independent and unofficial. It extends Codex with skills and hooks but is not Codex itself.
Discussion
No questions or reviews yet.
Related
PluginsA Claude Code skill that drives Codex over JSON-RPC: tasks, code reviews, parallel research and messaging between two agents.
An OpenAI Codex CLI extension: 27 skills, 16 MCP tools and AGENTS.md rules for PDCA development with a mandatory design document
A CLI that gives six coding agents project context, over 80 skills, verification and resumable tasks across sessions
A kit for porting a Claude Code workflow to OpenAI Codex CLI for multi-tenant PHP/Bitrix24-Marketplace projects
Codex Settings skills
Codex CLI Skills & Profiles
A Codex CLI plugin with twelve skills: deep research, pull request review, image generation and session handoff
OpenAI's official plugin: Codex reviews of your changes and task delegation to Codex from inside a Claude Code session
README in short
The README describes the `dev` skill family, the PABCD process, multi-model subagents, two-command installation, state settings and license credits. Korean and Chinese versions exist.