Agent Tutor Skill — AI skill for Claude Code

Agent Tutor Skill

A cognitive science-based AI tutor for [Claude Code](https://claude.ai/code) that won't let you fake understanding.

Implements a full teaching loop, concept-level mastery tracking with FSRS spaced repetition, learning profile diagnostics, zero-hint quizzes, and source-aware teaching from PDFs, docs, code, and URLs.

Why This Exists

AI makes learning feel easy. That's the problem.

Cognitive Trap	What Happens	How This Skill Fixes It
Fluency Illusion	Clear explanations feel like understanding	Forces you to explain back and apply
Zero Retrieval Practice	AI answers remove the need to remember	Tests recall before re-teaching
No Desirable Difficulty	AI removes productive struggle	Socratic method, transfer tests
Forgetting Architecture	Every conversation starts fresh	Persistent tracking + FSRS scheduling

Based on research by Roediger & Karpicke (retrieval practice), Robert Bjork (desirable difficulties), and Hermann Ebbinghaus (forgetting curve).

Install

npx skills add Bhala-Srinivash/agent-tutor-skill

Usage

/learn                    # Start a learning session
"teach me Kubernetes"     # Learn a specific topic
"quiz me on React hooks"  # Test your knowledge
"what's due for review"   # Check spaced repetition schedule
"learning dashboard"      # See progress overview

How It Works

Teaching Loop

Every concept follows: **Explain → Example → Check → Evaluate → Practice**

You can't say "I get it" without proving it. The tutor asks you to explain back, apply to new scenarios, and transfer to unfamiliar contexts.

FSRS Spaced Repetition

Unlike fixed intervals (3→7→14→30→90 days), FSRS adapts per concept:

• Easy concepts get longer breaks faster
• Hard concepts get reviewed more often
• Intervals personalize based on your recall history

Source-Aware Teaching

Point it at any source material:

• PDFs, docs, markdown files
• URLs and web pa