i-am-manware/Manware-s-AI-Learning-Toolkit

What it solves

This toolkit transforms AI assistants (like GitHub Copilot) from code generators into educational companions. It prevents the "copy-paste" trap where learners rely on AI to write their code, instead forcing the user to think, predict, and reason through problems to ensure actual learning occurs.

How it works

The project provides a set of structured prompts and instructions that configure an AI agent to act as a tutor. Instead of providing direct solutions, the AI is instructed to ask questions, provide incremental hints, and require the user to commit to an answer or hypothesis before moving forward.

It features 12 specific workflows (slash commands) for different learning scenarios, such as /hint for incremental problem solving, /debug for hypothesis-driven bug hunting, and /explain for verifying understanding through a "teach-back" test.

Who it’s for

Developers and students who want to use AI to learn programming concepts and architecture rather than just using it to generate code quickly.

Highlights

• Socratic Method: AI guides the user through questions and hints rather than providing the full solution.
• Diverse Learning Workflows: Specialized prompts for API exploration, architecture design, and post-mortem bug analysis (/autopsy).
• Adaptive Behavior: The AI adjusts its difficulty and questioning depth based on the user's performance.
• Learning Logs: An optional system for recording mistakes and concepts to enable spaced retrieval practice via the /retrieve prompt.
• Shipping Mode: A simple toggle ("ship this") to switch back to standard engineering mode when production speed is required.