Custom GPT to plugin migration: how OpenAI's process works

This page summarizes OpenAI's own documentation; see the sources below for the current wording.

Requirements

• You are the GPT's creator, or a workspace owner or admin.
• The GPT is published.
• Plugins are enabled for you.

Using someone else's GPT is not enough to migrate it.

The steps, per OpenAI

1. Go to My GPTs, find your GPT under Created by me and select Migrate to plugin.
2. Review the migrated skill and its reference files. Check that files and examples are present.
3. Rebuild any custom actions separately, using a supported app or a custom MCP server.
4. Test the plugin: right skill selected, instructions followed, harder case handled, needed tools available. Test both explicit selection (@ mention, or + then More) and a normal request.
5. Share it with the people or groups who need it. You need the Share plugins permission, and recipients install it unless an admin installed it for them.
6. Share the replacement link and update onboarding material and saved resources.

What a plugin is

OpenAI describes a plugin as an installable package that can contain skills (a folder with a SKILL.md file describing a repeatable workflow), an MCP server that connects to external systems, or both. See GPT vs plugin.

What does not transfer

Custom actions, the selected model, conversation starters (which may not copy one-to-one), previous chats and capability settings. Details in what transfers.

After you migrate

The original GPT stays usable until retirement but becomes read-only, so make edits in the plugin. OpenAI does not describe a way to export instructions or files from the plugin, so keep your own backup: instructions and files.