Currently if the model fails either because of no tokens left, not being initialised or any other reason - the agent silently fails instead of letting the user know.
Maybe respond to the user (or users preferred platform) letting them know that you are out of tokens or the model needs to be initialised.
Currently if the model fails either because of no tokens left, not being initialised or any other reason - the agent silently fails instead of letting the user know.
Maybe respond to the user (or users preferred platform) letting them know that you are out of tokens or the model needs to be initialised.