ndpvt-web/deeptutor-claude-skill
🛠️ Agent Skillndpvt-web
A Claude Code skill implementing graph-enhanced RAG for adaptive, dual-loop educational tutoring.
The deeptutor-claude-skill brings the sophisticated tutoring architecture of DeepTutor directly into the Claude Code environment. At its core, the project leverages graph-based Retrieval-Augmented Generation (RAG) to maintain a structured understanding of educational content. By utilizing NetworkX for graph manipulation, the skill allows the AI to traverse complex relationships between concepts, ensuring that tutoring sessions are grounded in accurate, interconnected knowledge.
Key technical features include a dual-loop solving mechanism that separates the reasoning process from the retrieval process, enhancing the tutor's ability to identify knowledge gaps. The skill supports the generation of structured research reports and adaptive questioning, which dynamically adjusts to the learner's progress. This implementation is specifically designed to bridge the gap between static document-based RAG and the nuanced, pedagogical requirements of personalized AI tutoring, making it a powerful extension for developers building educational AI agents.
💡Highlights
- ├─Graph-based RAG via NetworkX
- ├─Dual-loop solving architecture
- └─Adaptive pedagogical questioning
🎯For
- ├─AI Developers
- └─EdTech Engineers