Implement control mechanism for bad input

[Catrovitch]

• [x] Define "bad input"
• [x] Research how to control this (ideas: LLM checking input against example, LLM checking input to contain a certain level of detail, etc)
• [x] Frontend: Submit buttons for individual User input fields
• [x] Frontend: Notifications for pending/unsuccessful/successful validation of user input
• [x] Backend: Create prompts used for validation
• [x] Backend: Create backend endpoints for each user input that is to be validated

[Catrovitch]

Notes

Define bad input

• Use instructions given for each input to define what "good" input is

Ideas for control

• Prompt LLM with the Input-Instructions and Input and command LLM to decide whether the input follows the instructions.

Implementation

• Before each step-x implement a checking step that checks whether the input follows the instructions well enough.