HA7CH/ha7ch-school

What it solves

HA7CH School is an AI-native educational framework designed to move beyond static documentation. It transforms an AI agent (specifically for Claude Code) into a personalized tutor that guides users through courses on AI Native design and Forward Deployed Engineering (FDE), adapting the learning pace and order based on the user's existing skills.

How it works

The project is implemented as a "Claude Code skill," meaning it is a set of instructions and content that an AI agent loads to adopt a specific persona and knowledge base. It uses a progressive loading system where the agent only loads specific lesson files as the student progresses, keeping the context window clean. It also maintains a local state file to track learning progress across different sessions.

Who it’s for

• Aspiring AI Product Builders: Those wanting to learn "Zero Token" design and how to build AI-native products.
• Professional Services/Engineers: Individuals interested in the Forward Deployed Engineer (FDE) role, combining technical skills with business acumen.
• Non-Technical Users: People from business or operations backgrounds who need to learn GitHub as a collaboration language for the AI era.

Highlights

• Adaptive Learning: The tutor assesses the user's profile to dynamically reorder lessons, focusing more heavily on the user's weak points.
• Zero Token Design: The system updates itself by having the user's own agent perform the check and update, rather than using a centralized server to push updates.
• Zero-Simulation Training: Includes practical experiments where users create live pages or use built-in diagnostic tools.
• Cross-Session Memory: Tracks progress locally so students can resume learning exactly where they left off.
• Skill-in-Skill Architecture: Uses a modular structure where the agent loads specific lessons only when needed to optimize performance.