OpenAI Plugins Example Repository

Key takeaways#

• openai/plugins is a curated repository of Codex plugin examples.
• It should be treated as a resource and example collection, not a single installable tool.
• Each plugin example lives under plugins/<name>/ and uses a .codex-plugin/plugin.json manifest.
• The repository includes examples for workflows such as Figma, Notion, web apps, iOS, macOS, Expo, Netlify, Remotion, and Google Slides.

What is openai/plugins?#

openai/plugins is OpenAI's public example repository for Codex plugins. The repository demonstrates how plugin bundles can package reusable workflows for Codex. It is not one product with one install command. Instead, it is a source-backed reference library that developers can inspect when they want to understand how Codex plugins, skills, app integrations, commands, hooks, and MCP-backed surfaces can be organized.

The README states that each plugin lives under plugins/<name>/ and has a required .codex-plugin/plugin.json manifest. Optional companion surfaces can include skills/, .app.json, .mcp.json, plugin-level agents/, commands/, hooks.json, assets/, and other supporting files. That structure makes the repository useful as a pattern library for teams building their own Codex workflows.

What examples are included?#

The repository has examples for product and engineering workflows. Highlighted plugin directories include Figma, Notion, build-ios-apps, build-macos-apps, build-web-apps, Expo, Netlify, Remotion, and Google Slides. These examples show how a plugin can encode domain-specific instructions, assets, integrations, and workflow hooks so Codex can work with less repeated setup.

Who should use it#

This resource is useful for developers building internal Codex workflows, platform teams evaluating plugin structure, and AI engineering teams that want examples before designing their own plugin bundles. It is also a good reference for comparing where to put instructions, manifests, command definitions, MCP configuration, and supporting assets.

Important caveats#

Do not treat this repository as a single marketplace app or a general-purpose CLI. Individual plugin examples may be usable or installable in the right Codex environment, but the repository itself is an examples collection. GitHub metadata checked during this run showed about 2,645 stars and 310 forks. The GitHub API did not report a license, so license-sensitive users should inspect the repository before reusing code in production.

Why it matters for builders#

Agent workflows become easier to maintain when repeatable context is packaged instead of pasted into every prompt. The openai/plugins repository shows that packaging layer in concrete form. Builders can study the manifests and directory layout to design cleaner Codex workflows for design systems, app development, documentation, deployment, and team-specific operations.