The goal is to implement an AI agent that generates praise messages in a ‘broken tone’—adopting the format of system error logs or patch notes—based on short text input by the user (today's activities, emotions, self-criticism, to-do notes, etc.). Based on simple user feedback regarding the result (e.g., "Really?" / "Not good," etc.), the agent reflects this feedback to adjust the humor intensity, sincerity tone, sentence length, and bug type of the subsequent response.

The goal is to implement an AI agent that generates praise messages in a ‘broken tone’—adopting the format of system error logs or patch notes—based on short text input by the user (today's activities, emotions, self-criticism, to-do notes, etc.). Based on simple user feedback regarding the result (e.g., "Really?" / "Not good," etc.), the agent reflects this feedback to adjust the humor intensity, sincerity tone, sentence length, and bug type of the subsequent response.