OpenAI Rejecting Code Execution Plugins for no reason

OpenAI raise this to security team for 2 week and they decides to reject the plugin.

They don’t even give proper reason for this.

Plugin Info:
Plugin Name : Code Runner.
Submission : Re-submission
Plugin time : 2 Weeks ( 1 Week for security issues).
Plugin Description : Lets you run and manage your code in 70+ coding languages.

If you had similar experience then please share with us.

@logankilpatrick resolved this ticket now so no issues are there now
Thanks logan.

The reason is right there.

Your plugin executes code → not approved.


did you read it properly it was already approved , it was re-submission with just Plugin URL changed thats it, how can they not approve it when its already in Plugin store?

It may have been erroneously approved before and they caught the error on resubmission or they decided to more strictly enforce the rule in the interim.

1 Like

There are similar plugins for this as well - approved
CodeCast WanderBox - Executes Code - Approved.
CoderPad - Execute code with online IDE - Approved.

There could be any number of reasons why they would approve one plugin but not another.

The simplest answer is your plugin runs code and they aren’t approving plugins which run code at this time, no matter how unfulfilling they may be.

Hey Haseeb,

I can think of a couple of possibilities. They might be worried that you are competing directly with their code interpreter, which they had been selectively opening up to people. While I’m not a lawyer, I imagine this one is less likely considering that sounds pretty much like the same anti competitive practices that Microsoft was sued for back a while back.

Another possible reason is that they have some serious concerns about AI with access to code execution and yours crosses a threshold they are uncomfortable with. And the last one I think could be plausible is that, like most environments that can execute user provided code, they are concerned with legal aspects of software from their side generating code that can be executed on your devices.

Opens up a lot of legal problems for them.

As for your examples, CodeCast only mentions “compiles your code” they don’t mention execute. I haven’t used it though. CoderPad completes the code for you in an IDE, but I haven’t seen anywhere that says it automatically executes the solution.

The two existing ones make me think that maybe they want a human in the loop to overview and execute the code manually. I haven’t used any of those though, so I could be off the mark. It could be they are comfortable executing code on their side automatically, but not other people doing that as a product/service. Something to consider.

Maybe you can find a similar solution/compromise with your product/service.

1 Like

I’m rereading the message from them, and it sounds like they might have an automatic GPT solution rejecting plugins. Maybe I’m crazy and been talking to GPT too long, but the “we are currently unable to approve any plugins associated with managing and executing code” sounds very GPT like. It has a tendency of writing “whole sentences” like in grade school.

So it might be an automated pipeline responsible for this and there is some human just in the background approving rejections.

My plugin is unique because it can generate graphs :bar_chart: and can manage documents as well for your code , so they mentioned management aswell and i also changed title description to “Run your code” rather than “Compile your code”.

That could be the reason.

1 Like

The Noteable plugin also allows for the execution of AI generated code in notebook. Noteable was also off the Plugin Store for a short time, but they say it is back now.

We have been working a lot on these plugins, we deserve at least good communication, some insight, real reasons and a way to resolve the situation in a way which is good for both OpenAI and the plugin developers.

Quoting from:

1 Like

Yes i agree they need to give proper reason to developer of plugins what they did wrong and what could be improved, it seems like every plugin with code execution or code analysis plugins has been affected

There was some confusion with this Code execution plugin but not it was resolved by Logan.

1 Like

Are you able to elaborate? Maybe your situation can help someone in the future. Did they just not understand what the plugin was doing?

I just DM to Logan on Twitter and he resolved the ticket saying “The issue has been resolved” thats it.

1 Like