Neurospicy explanation
A plugin packages related capabilities so you can install them together in supported ChatGPT and Codex experiences. It might contain a few skills, a connection to another service, or both. A skill describes a job; a plugin is a way to bundle and share it. Some plugins request access to tools or accounts. Check what is included before installing one.
// Put it to use
A real-world example
You have three useful skills for project notes and want to keep them together rather than explain how to copy three separate folders.
- Check that each skill works on its own: tidy notes, extract actions and draft a follow-up.
- Use the official plugin guide to package them as a skills-only plugin.
- Test the package locally with sample notes and inspect the output from each skill.
- Share it only when its contents and instructions are clear; an external service connection can be added later if there is a real need.
What you getOne installable package for a related set of workflows, without pretending that every workflow needs a new account connection.
// The useful bits. The trade-offs.
Is it a good fit?
People ready to organise or share a tested skill.
Pros
- Keep related skills together in a single package.
- Start with written workflows and add supported connections only when needed.
Cons & considerations
- Supported features depend on the host and execution environment.
- Installation, connector access and public distribution are separate actions.
A little more detail
Understand the files in a small skills-only plugin.
A plugin can distribute reusable skills and optional connections. A minimal skills-only package contains a plugin.json manifest and a skills directory.
Begin with one clearly scoped skill. Read the complete package and its dependencies before installing it. Packaging instructions does not establish that connectors are authenticated or hooks are trusted.
// Take the idea further
What else could you do?
A writing toolkit
Package separate skills for drafting, checking facts and reviewing tone.
A project starter kit
Bundle a small set of reviewed project workflows, then test each with sample inputs.
References & credits2 sources +
OpenAI documentation team
Sources checked . Features and terms can change.