awesome-codex

mcp
Security Audit
Warn
Health Warn
  • License — License: CC0-1.0
  • Description — Repository has a description
  • Active repo — Last push 0 days ago
  • Low visibility — Only 5 GitHub stars
Code Pass
  • Code scan — Scanned 2 files during light audit, no dangerous patterns found
Permissions Pass
  • Permissions — No dangerous permissions requested

No AI report is available for this listing yet.

SUMMARY

A curated list of awesome tools, skills, MCP servers, and AGENTS.md resources for OpenAI Codex. Updated weekly.

README.md

Awesome Codex Awesome

English | 简体中文

A curated list of tools, skills, MCP servers, AGENTS.md patterns, and resources for OpenAI Codex.

Codex is OpenAI's agentic coding agent — a terminal CLI, an IDE extension, and a cloud environment that share the same agent. This list focuses on the community ecosystem around it, plus the official docs worth bookmarking.

What makes this list different: it is updated every week, and entries are kept alive. Dead links get removed, not archived. If a section stops being useful, it gets deleted. See OPERATIONS.md for the maintenance contract.

Legend: (Official) by OpenAI · (Chinese) Chinese-language resource · (Paid) paid or freemium · everything else is community/open source.

Contents


Official Resources

  • openai/codex — The Codex CLI itself. Rust, Apache-2.0, developed in the open. (Official)
  • Codex Documentation — Entry point covering CLI, IDE, and cloud surfaces. (Official)
  • Codex Changelog — Release notes. Codex ships fast; this is the single highest-value page to check weekly. (Official)
  • Codex CLI Reference — Commands, flags, and terminal workflow. (Official)
  • Security & Approvals — Sandbox modes and the approval model. Read this before granting write access. (Official)
  • Prompting Guide — OpenAI's own guidance on prompting Codex. (Official)
  • openai/skills — Official skills catalog for Codex. (Official)

Getting Started

  • Quickstart — Install, sign in, and run the first task. (Official)
  • openai/codex — Releases — Prebuilt binaries and per-release notes. (Official)
  • openai/codex — Discussions — Where the maintainers answer questions and the community posts ecosystem projects. (Official)
  • ChatGPT desktop app for Linux — Install the preview desktop app on supported Ubuntu, Debian, and Fedora distributions to work with local files and Codex. (Official)
  • Import from another agent — Bring supported setup, skills, plugins, projects, and recent work from Claude Code or Cursor into ChatGPT or Codex. (Official)

AGENTS.md

AGENTS.md is the file Codex reads before doing any work — the equivalent of a README written for the agent. It layers: ~/.codex/AGENTS.md (global) → repo root → nested directories.

Configuration

  • Config File Reference~/.codex/config.toml: models, approval policy, sandbox, MCP servers, profiles. (Official)
  • Codex Models — Choose models and reasoning effort in interactive and non-interactive Codex sessions. (Official)
  • Auto-review — Routes sandbox-boundary approval requests to a separate reviewer agent without expanding the active sandbox or permission policy. (Official)
  • CLI Customization — Customize the Codex terminal interface, editor, completions, and keyboard shortcuts. (Official)
  • Codex Rules: Global Instructions, AGENTS.md, and the Mac App — Walkthrough of how the instruction layers actually resolve in practice.
  • farion1231/cc-switch — Cross-platform desktop app for switching provider/account configs across Codex, Claude Code, and others.

Skills

Skills package reusable instructions and scripts that Codex can load on demand.

  • Build Skills — Official guide to authoring skills. (Official)
  • openai/skills — OpenAI's skills catalog. Start here before writing your own. (Official)
  • Record & Replay — Turns a demonstrated macOS workflow into a reusable Skill; it requires Computer Use and is not available in every region. (Official)
  • Suede Creator Skills — MIT-licensed Codex and Claude Code skills for orchestration, worker fleets, code review and ship gates, AI evaluation, product, design, and growth.
  • composio-community/awesome-codex-skills — Curated practical skills for the Codex CLI and API.
  • VoltAgent/awesome-agent-skills — Large cross-harness skill collection; Codex-compatible subset is substantial.
  • Nutlope/hallmark — Design skill that pushes generated UI away from generic AI-template output. Works with Codex.
  • microsoft/skills — Skills, MCP servers, and AGENTS.md files for grounding coding agents in Microsoft SDKs.

Plugins

Plugins package reusable Codex capabilities for installation and sharing.

  • Plugins — Browse, install, and use reusable capability bundles in the Codex CLI and supported ChatGPT surfaces. (Official)
  • Build Plugins — Package skills and MCP servers into installable bundles, then test them through a local marketplace. (Official)
  • Skills & Plugins — Decide when a repeatable workflow should remain a skill and when it should become a shareable plugin. (Official)
  • Product Design — Codex plugin for synthesizing user research, creating interactive prototypes from briefs or visual references, and reviewing user flows to guide design iteration. (Official)
  • nowork-studio/notfair-plugin — Codex marketing plugin with 45 SEO, GEO, paid-media, and analytics workflows plus one hosted MCP connection for live account work.
  • MARGINAL — Local-first Codex runtime governor that observes repeated tool work in Shadow Mode and gates narrow enforcement on verified evidence, explicit consent, and integrity checks.

MCP Servers

Codex speaks the Model Context Protocol, so most MCP servers work without Codex-specific glue.

Hooks & Automation

  • Hooks — Run commands on Codex lifecycle events. (Official)
  • openai/codex-action — Run codex exec in GitHub Actions with explicit API credentials and permission profiles. (Official)
  • caliber-ai-org/ai-setup — Syncs agent skills, MCP servers, and config across machines with one command.
  • mksglu/context-mode — Sandboxes tool output to cut context usage; supports Codex.

Codex SDK & Headless

  • Codex SDK — Embed Codex in your own applications and pipelines. (Official)
  • Codex App Server — Exposes authentication, conversation history, approvals, and streamed agent events over JSON-RPC for building rich Codex clients. (Official)
  • codex exec — Non-interactive Mode — Run Codex in scripts and CI with pipe-friendly output and explicit sandbox settings. (Official)

IDE & Editor Integrations

Cloud & Code Review

  • Codex Cloud — Delegate tasks to cloud containers that open PRs. (Official)
  • Code Review — Codex reviewing pull requests. (Official)
  • Review GitHub pull requests with Codex — Set up Code Review for GitHub, request reviews with @codex review, and write custom review rules in AGENTS.md. (Official)

Multi-Agent & Orchestration

  • Subagents — Run independent work in parallel and define specialized agents with task-specific models and instructions. (Official)
  • Git Worktrees — Isolates parallel Codex desktop chats in managed Git worktrees so they can work in the same repository without interfering with each other. (Official)
  • Local Environments — Configure common actions and setup scripts for Codex worktrees. (Official)
  • Long-running Work — Keep multi-step work focused with clear outcomes and completion criteria. (Official)
  • wshobson/agents — Multi-harness agentic plugin marketplace covering Codex CLI, Claude Code, Cursor, and more.
  • AgentWrapper/agent-orchestrator — Agent IDE for managing fleets of coding agents.
  • code-yeongyu/oh-my-openagent — Opinionated harness (omo / lazycodex) built for long, complex agent runs.
  • ruvnet/ruflo — Meta-harness for coordinating multi-agent swarms.
  • YYLO — Command-line orchestrator for coding agents: each task runs in a dedicated Git worktree with typed validation and a risk-based merge queue, and Codex is one of its built-in subagent runtimes.

Clients & Remote Access

  • slopus/happy — Mobile and web client for Codex and Claude Code, with realtime voice and end-to-end encryption.
  • siteboon/claudecodeui — Web and mobile UI for driving Codex, Claude Code, Cursor CLI, and OpenCode.
  • chenhg5/cc-connect — Bridges local coding agents to messaging platforms so you can drive them from chat.
  • iOfficeAI/AionUi — Local, open-source desktop app for running Codex and other agents continuously.

Tools & Utilities

  • Graphify-Labs/graphify — Turns a codebase plus its docs, schemas, and PDFs into a queryable knowledge graph for agents.
  • colbymchenry/codegraph — Pre-indexed code knowledge graph that auto-syncs on change; supports Codex.
  • OthmanAdi/planning-with-files — File-based persistent planning for long-running agent tasks.
  • Skill Auditor — Security scanner for Codex and AI agent skills that detects prompt injection, malicious scripts, credential access, data exfiltration, and supply-chain risks.
  • Skill Sunset — Local, read-only audit for stale AGENTS.md, CLAUDE.md, and generic SKILL.md instructions.
  • nexu-io/open-design — Local-first desktop app that gives a coding agent a design surface to work against.
  • google-labs-code/design.mdDESIGN.md spec: describes a visual identity to coding agents, the way AGENTS.md describes a codebase.

Chinese Resources

  • liyupi/ai-guide — 程序员鱼皮的 AI 资源大全与 Vibe Coding 教程,含 Codex 在内的多种 agent 用法。 (Chinese)
  • jlcodes99/cockpit-tools — 通用 AI IDE 账号管理工具,支持 Codex、Copilot、Cursor 等。 (Chinese)
  • Wei-Shaw/claude-relay-service — 自建中转服务,统一接入 Claude、OpenAI 等订阅,供本地 agent 使用。 (Chinese)

Related Awesome Lists

Listed honestly, including the ones that compete with this one. Cross-check them — they cover different slices.

Community

Contributing

Pull requests are welcome. Read CONTRIBUTING.md first — the quality bar is stricter than "it's a link about Codex."

Weekly changes are recorded in CHANGELOG.md.

License

CC0-1.0 — public domain.

Reviews (0)

No results found