openai/plugins

OpenAI Plugins

What this project is

openai/plugins is a collection of example plugins for Codex — OpenAI's coding assistant. Each plugin is a small add-on that gives Codex extra abilities or guidance for a specific tool or workflow.

Think of a plugin like a "skill pack" you can drop into Codex so it knows how to work with Figma, Notion, iOS apps, web apps, and more.

How plugins are organized

Every plugin lives in its own folder under plugins/<name>/. Inside that folder, the only required file is a manifest called plugin.json (located at .codex-plugin/plugin.json). This file describes what the plugin is and how Codex should load it.

Plugins can also include optional extras:

  • skills/ — step-by-step instructions Codex can follow
  • .app.json — app configuration
  • .mcp.json — settings for connecting to external tools via the Model Context Protocol (a standard way for AI tools to talk to services)
  • agents/, commands/, hooks.json, assets/ — additional behaviors, shortcuts, event triggers, and static files

How plugins are distributed

There's a marketplace — a catalog file that lists available plugins. The default one is at .agents/plugins/marketplace.json and points to the standard plugins/ folder. Users who log in with an API key get a separate catalog at .agents/plugins/api_marketplace.json.

Highlighted examples

The README calls out several richer plugins:

  • plugins/figma — helps Codex use Figma: reading designs (use_figma), converting designs to code (Code to Canvas), syncing code with design (Code Connect), and applying design-system rules
  • plugins/notion — helps with planning, research, meetings, and capturing knowledge in Notion
  • plugins/build-ios-apps — guidance for building SwiftUI apps: implementation, refactoring, performance tuning, and debugging
  • plugins/build-macos-apps — macOS-specific SwiftUI/AppKit workflows, including build/run/debug loops and packaging
  • plugins/build-web-apps — web development workflows: deployment, UI, payments, and databases
  • plugins/expo — for Expo and React Native apps: SDK upgrades, EAS (Expo's build service) workflows, and Codex Run actions
  • plugins/netlify, plugins/remotion, and plugins/google-slides — additional public bundles that combine skills and MCP-backed integrations

Bottom line

This repo is a showcase and starting point — it demonstrates how to structure Codex plugins, what files they need, and what kinds of workflows they can enable. If you wanted to build your own plugin, this is the reference to copy from.

Related

  • Project
    tailcallhq/forgecode

    An AI-enhanced terminal development environment that integrates coding agents directly into the shell via a TUI, CLI, or ZSH plugin.

  • Project
    XiaoDuoYa/codex-with-chatgpt

    Codex with ChatGPT is a bridge that lets the official ChatGPT web UI act as the planning/review brain for a Codex coding agent, saving API quota. It creates a secure, read‑only MCP connection (OAuth‑protected, Cloudflare‑tunneled) so ChatGPT can read only the needed files, while Codex executes code locally. Install via a one‑paste script, copy the provided Skill into your Codex skill folder, run `c2c setup`, and then command Codex to use the skill. The project is an unofficial, MIT‑licensed Node.js tool that focuses on security (read‑only, path containment, secret‑file denial) and works without an OpenAI API key.

  • Project
    anthropics/defending-code-reference-harness

    A reference implementation for autonomous vulnerability discovery and remediation that uses Claude to find, verify, and patch security bugs in source code.

  • Project
    algorithmicsuperintelligence/openevolve

    OpenEvolve is an open‑source Python library that combines large‑language‑model prompting with a MAP‑Elites evolutionary algorithm (island parallelism) to automatically rewrite, optimize, or invent code across many languages. Users supply an initial program and an evaluator; the system mutates the code via LLMs, evaluates candidates, and maintains a diverse Pareto‑optimal archive. It supports multiple LLM back‑ends (OpenAI, Gemini, Claude, local Ollama, etc.), offers deterministic reproducibility, extensible hooks, and can be run via CLI, Docker, or as a Python module. Example use‑cases include GPU kernel tuning, novel algorithm discovery (e.g., adaptive sorting, circle packing), scientific computing, and competitive‑programming solution generation.