CourseBuilder — AI Learning Platform

An AI-powered course builder. Describe a course in chat, the AI asks a few clarifying questions and drafts a structured plan, then generates each section (with an automated critic review), per-module quizzes, and a final capstone assignment — all reviewed and approved by you.

There are two ways to use it:

• Web app — FastAPI backend (api.py) + React/Vite frontend (frontend/)
• CLI — a terminal REPL (main.py)

---

Prerequisites

• Python 3.10+
• Node.js 18+ and npm
• An API key for one LLM provider: Gemini, OpenAI, Claude, or Groq

---

1. Backend setup

cd learning-platform

# create + activate a virtual environment
python -m venv venv
source venv/bin/activate          # Windows: venv\Scripts\activate

# install dependencies
pip install -r requirements.txt

Configure the .env

Create a file named .env in the project root (you can copy .env.example):

cp .env.example .env

Set LLM_PROVIDER to the provider you want, and add only that provider's key. Pick one of the following:

# Option A — Google Gemini
LLM_PROVIDER=gemini
GEMINI_API_KEY=your_gemini_api_key_here

# Option B — OpenAI
# LLM_PROVIDER=openai
# OPENAI_API_KEY=your_openai_api_key_here

# Option C — Anthropic Claude
# LLM_PROVIDER=claude
# ANTHROPIC_API_KEY=your_anthropic_api_key_here

# Option D — Groq
# LLM_PROVIDER=groq
# GROQ_API_KEY=your_groq_api_key_here

Optional — pin a specific model (otherwise a sensible default per provider is used):

# LLM_MODEL=gemini-2.5-flash-lite

Provider	LLM_PROVIDER	Key variable	Default model
Gemini	gemini	GEMINI_API_KEY	gemini-2.5-flash-lite
OpenAI	openai	OPENAI_API_KEY	gpt-4o
Claude	claude	ANTHROPIC_API_KEY	claude-haiku-4-5
Groq	groq	GROQ_API_KEY	llama-3.3-70b-versatile

.env is git-ignored — your keys never get committed.

Run the API server

source venv/bin/activate
uvicorn api:app --port 8000 --reload

The API runs at http://localhost:8000.

---

2. Frontend setup

In a second terminal:

cd learning-platform/frontend

# install dependencies
npm install

# start the dev server
npm run dev

Then open http://localhost:5173/course-builder.

The Vite dev server proxies all /session API calls to the backend on http://localhost:8000, so just keep both terminals running. Make sure the backend is up before using the app.

To build for production: npm run build (output in frontend/dist/).

---

3. CLI (optional)

Prefer the terminal? With the backend .env configured and the venv active:

source venv/bin/activate
python main.py

Commands: /upload <path>, /status, /export, /ppt <section-id>, /reset, /help, /quit. Keywords: approve plan, approve section, revise: <note>.

---

How it works

CLARIFYING → PLAN_DRAFTED → CONTENT → ASSESSMENT → DONE

1. Chat — describe your course; attach reference files (PDF, DOCX, PPTX, TXT) for the AI to use as source material.
2. Plan — the AI drafts a structured plan (max 4 modules × 2 submodules). Review it on the right and approve, or keep refining.
3. Content — each section is generated and critic-reviewed; you approve or revise it.
4. Assessments — generate one quiz per module plus a capstone final assignment.
5. Export — download the full course (plan + sections + assessments) as JSON.

---

Project layout

learning-platform/
├── api.py            # FastAPI web backend
├── main.py           # CLI REPL
├── agent.py          # planning conversation agent
├── generator.py      # section / quiz / assignment generation
├── prompts.py        # system prompts, schemas, limits
├── store.py          # in-memory plan + section state
├── models.py         # data models
├── llm.py / config.py# provider-agnostic LLM factory
├── file_extractor.py # PDF/DOCX/PPTX/TXT text extraction
├── requirements.txt
└── frontend/         # React + Vite + Tailwind UI