If you’ve connected OAuth, and if you step away from a chat session for too long – perhaps if you’ve been backed off for usage limits – OAuth won’t ever resume for a set of actions. This will send your LLM into an endless circle of retrying actions.
Prematurely ending chats often means a failed “training” session with the LLM.
May we PLEASE have some solid middleware documentation, and does it include a method to force, re-trigger, or clear OAuth? Or can nobody log out?
Can our actions be called from Python? Your promoted “ActionsGPT” gives me two methods that both fail, despite being specifically written ON that platform, lol. And where is this documented?