Personal Lesson Agent

An AI-powered desktop learning platform (React 19, TypeScript, Electron) that turns a short interview into a complete personalized course — lessons, multi-file coding exercises, and quizzes — with a context-aware tutor beside every module.

It’s built on a custom multi-agent framework: four specialized agents (Interview, CourseDesigner, ContentGenerator, Tutor) extend a shared abstract base with a bounded tool-use loop, calling Claude through a typed OpenRouter client.

The defining design is a four-level hierarchical context model — User → Subject → Course → Module — so every AI interaction knows the learner’s background, tools, and current objective. Data persists as inspectable JSON and Markdown on disk, the Electron renderer is sandboxed behind a path-validated IPC layer, and exercises open directly in the user’s IDE.