sugarlabs/reflection-engine ? reverse-engineered prompt

Reverse engineered prompt

Build me a small Python library for short reflection conversations with children about something they just made.

It should ask a few simple questions like what took the longest, what changed, and what they might try next time. It should never praise them, grade them, or answer for them, just keep the conversation focused on the work. I want the library to be stateless, with one model call per turn, and it should return a typed result that the host app can display and store. The host should handle all the child facing text and the conversation history, so keep that out of the library itself.

Please make it follow the project spec and tests, and keep the design clean enough that it could be used by something like a journal app through a provider that talks to a model. If you need to check current model API details, look up the docs online.

Are you gonna build this?

make sure you review the code using coderabbit

Try freeSponsored — opens CodeRabbit in a new tab