Thanks for the clarification! Prodigy is very centered around collecting training data for machine learning models, which should typically be as structured as possible. So wherever possible, it tries to discourage you from collecting free-form input (which is often difficult to evaluate and compare).
However, there are always exceptions to that rule and Prodigy does support fully custom templates with HTML, CSS and JavaScript. So you could add one with input fields that then update the current task with whatever the user has typed into the field. This should be fairly straightforward to build. You can find examples and more details here:
You could also experiment with a combination where you let the annotators highlight spans in the text and ask for additional information. But that depends on the specifics of your task.