I’ve got an automation which calls an AIP agent based action. The issue I’m having right now is that the automation runs with the permissions of whoever last saved the automation, this causes us some real issues as the behaviour can change purely on who last edited the automation.
I see there is an option to run project scoped automation but unfortunately this doesn’t support AIP agents based functions. I had an idea too create a function compute module with a service user that has a function that purely passes through the users query to run the AIP agent with the service users permission, but this feel quite clunky. Is there any easier way to achieve this without having to spin up a module for each agent?
Yes unfortunately AIP Agent functions are only supported in the User scoped security mode for now - it is on our roadmap to add support for Project scoped mode, but no confirmed timeline on this feature yet.
To allow you to run your agent with application permissions instead for now, I’d suggest something similar to your idea of creating a Developer Console application and using its service user to call the agent through the platform APIs.
You can then set up a Data Connection source with your service user credentials in order to use this for platform API calls in a Function (docs).
An alternative to the source + Function approach is to use the Compute Module approach, as you suggested.