Understand the code you write — especially when AI wrote it.
Learning Layer watches the files you save, detects the programming concepts your
code uses, tracks what you already know, and explains the worthwhile ones in
plain language, referencing your actual code. Built for bootcamp students,
junior devs, and anyone "vibe coding" faster than they can keep up with.
Features
Teach on save — save a .js/.ts/.jsx/.tsx file and get a lesson card for
the concepts worth learning, with your code highlighted.
Concept board — a sidebar (🎓) and dashboard showing your concepts grouped
by area and mastery (Understood / Practicing / Shaky / New).
Bring your own model — Anthropic (Claude) or any OpenAI-compatible endpoint
(OpenAI, Ollama, OpenRouter, Groq, LM Studio…). Your key is stored in VSCode
SecretStorage; nothing is sent anywhere except your chosen provider.
Getting started
Run “Learning Layer: Configure Model (API Key)” from the Command Palette.
Anthropic → model claude-opus-4-8, your sk-ant-… key.
OpenAI → model gpt-4o-mini, your sk-… key.
Ollama (free, local) → base URL http://localhost:11434/v1, model
llama3.1, no key.
Run “Learning Layer: Test Model Connection” to confirm it works.
Open a JS/TS/React file and save. A lesson appears; the 🎓 icon in the
activity bar shows your board.
Without a configured model, Learning Layer still works using built-in
explanations — you just don't get ones tailored to your specific code.
Settings
learningLayer.teachOnSave — analyze on every save (default on).
learningLayer.maxPerSession — how many concepts to surface at once.
learningLayer.minThreshold — minimum teaching-value score to surface.
learningLayer.provider / model / baseURL — your model setup.
