Hi everyone, my current understanding regarding Pydantic return types for tasks is that it validates it but it doesn't provide the schema of the Pydantic model in the prompt. This is based on what I can see in this task.jinja prompt template and trying things out. It simply prints out the name of the model class e.g. for a Pydantic model

User

it would be

result type: <class '__main__.User'>

. Two questions: 1. First of all, is the above correct? If so, would it not be better to provide the json schema for Pydantic models with something like

User.model_json_schema()

? 2. General question, is there any plan to make the prompt templates customisable or do you see this as a benefit, having it fixed and only tweakable through the entity descriptions/instructions etc.? a. One thing I think I would have appreciated is the documentation showing these templates. It makes it very tangible how the different components work together tasks/agents/tools etc.