Skip to content

Latest commit

 

History

History
48 lines (37 loc) · 2.36 KB

01-research-context.md

File metadata and controls

48 lines (37 loc) · 2.36 KB

Research Context

Origin

Charles Gouilliard wants to learn Python for summer 2026 and his coding courses starting next year. Rather than using a generic course or AI chatbot, the goal is to build a Claude Code skill that acts as a personalized, persistent tutor — combining the best of AI tutoring with real developer workflows.

Landscape Analysis

Existing Solutions Reviewed

Platform Strengths Weaknesses
PySchool.ai AI hints, self-paced Web-only, no memory across sessions
Companion AI VS Code integration, free Limited curriculum structure
BootSelf AI Session memory, mentorship Web-based, not file-based coding
Replit AI Inline suggestions, full projects Writes code FOR you — bad for learning
Algocademy Adaptive challenges Focused on algorithms, not general Python
Codecademy Interactive basics, free tier No personalization, no memory
Coursera (Python for Everybody) Structured, university-quality Static, no AI adaptation
ChatGPT as tutor Conversational No memory, no file-based coding, gives answers too easily

What's Missing Everywhere

  1. Persistent memory — No tool remembers what you struggled with last week
  2. Real files — Most tools use web editors, not real dev workflows
  3. Socratic method — Most AI gives the answer instead of teaching
  4. Adaptive difficulty — Courses are linear, not responsive to the learner
  5. Spaced repetition — No coding tool uses proven memory science
  6. Integrated testing — Learner doesn't get automated feedback on correctness

Why Build a Claude Code Skill

  • Claude Code already runs in the terminal alongside VS Code
  • Skills system supports persistent file-based memory
  • Can execute Python, run tests, read user code
  • User (Edouard) already uses Claude Code daily
  • Free, private, fully customizable
  • Charles learns in a real developer environment from day 1

References