I’m interested in these answers too - we are getting ready to submit our plugin to the store but anticipate adding features over time and the docs aren’t super clear. At https://platform.openai.com/docs/plugins/production/updating-your-plugin it mentions
Manifest files must be manually updated by going through the “Develop your own plugin” flow in the plugin store each time you make a change to the
ai-plugin.json
file. ChatGPT will automatically fetch the latest OpenAPI spec each time a request is made so changes made will propagate to end users immediately. If your plugin is available in the ChatGPT plugin store and you go through the “Develop your own plugin” flow, we will automatically look for changes in the files and remove the plugin if it has changed. You will have to resubmit your plugin to be included in the store again.
I think this says that the process to update a plugin is:
- Modify plugin, making sure to modify something in the
ai-plugin.json
manifest so the system can pick up that there’s been a change [and/or the openapi spec? see 3 below] - Use the “develop your own plugin” again, which will cause OpenAI to fetch the new spec for the plugin
- Since the
ai-plugin.json
has changed (maybe it also looks atopenapi.json
since it looks for “changes in the files”), it will remove the plugin from the store. I wonder what - At which point we can use the resubmission process, once approved it will be re-added
So it sounds like the plugin will have to be offline while the new submission is reviewed?
I also wonder how it knows which plugin to disable in the store, maybe by matching the name_for_model
or name_for_human
? Plugins aren’t versioned, I saw a thread somewhere that folks were putting version numbers in the name_for_model
during local development.
related: