1786329860/deep-teach

Deep-teach is a prompt protocol designed to transform how AI coding assistants help people learn programming. The core problem it addresses is that when an AI writes code for you, you get a result but you do not understand what was done or why. Deep-teach changes this by making the AI automatically produce a structured "Teaching Card" after every coding action, so users learn the reasoning behind each decision rather than just receiving output. The README is written in Chinese. The concept is that after each coding operation, whether writing a function, installing a dependency, designing a database table, or fixing a bug, the AI outputs a six-part analysis card covering: what technology was used and its role, why that technology was chosen over alternatives with a trade-off comparison, a deep technical explanation of the underlying principle, a comparison of at least two alternative approaches, a summary of key strengths with supporting data, and guidance on transferring the knowledge to other domains. The system has three card variants: a full standard card for decisions with real architectural weight, a short mini card for trivial actions like adding a console log, and an enhanced card for major architectural choices that also includes a decision tree and risk matrix. It is loaded as a custom skill into AI coding tools like Cursor or Claude Code. The target users are non-computer-science students using AI for coursework, self-taught developers, and junior engineers who want to genuinely understand technology choices rather than just copy-paste solutions.