Exam Simulator
WAOmaster/Exam-simulator
Simulador de exames com IA que diagnostica e corrige lacunas de aprendizado
Ver no GitHubPlataforma de aprendizado adaptativo que transforma qualquer conteúdo em exames interativos. Usa Google Gemini 3 com diagnóstico cognitivo profundo para identificar lacunas de conhecimento, gerar explicações visuais dinâmicas e oferecer feedback personalizado através do método socrático.
Instalação
npx convez-ai WAOmaster/Exam-simulator1 comando no terminal: detecta o formato do repositório (marketplace ou plugin avulso) e instala via Claude Code. Requer Node 18+ e o Claude Code instalado.
Métricas
Casos de uso
- Preparação para certificações técnicas (AWS, Cloud, etc)
- Estudos acadêmicos com análise de conceitos mal compreendidos
- Treino para entrevistas técnicas e LeetCode
- Análise visual de problemas matemáticos e científicos via upload de imagens
Veredito

Ferramenta madura e bem-arquitetada em TypeScript/Next.js que vai além de quizzes simples ao diagnosticar *por que* o usuário erra. A integração com Gemini 3 Pro (thinking_level alto) e geração dinâmica de visualizações Python é diferencial legítimo, embora exija ambiente com dependências pesadas. Recomendado para educadores e desenvolvedores que precisam de feedback diagnóstico real.
README
A tradução para o português deste README ainda está na fila.
ExamSimulator 🎓
AI-Powered Adaptive Learning & Cognitive Diagnostics
ExamSimulator is a next-generation learning platform that transforms any content into an interactive exam experience. Powered by Google Gemini 3, it goes beyond simple quizzing by acting as a Cognitive Companion—diagnosing why you get questions wrong and dynamically generating visual explanations to fix your knowledge gaps.
🧠 What Makes This Different?
Most exam tools are binary: you are either "Right" or "Wrong." ExamSimulator understands the student, not just the subject.
- Universal Subject Support: Whether you are studying for AWS certifications, high school biology, or LeetCode interviews, ExamSimulator adapts to your material.
- Deep Think Diagnostics: Uses Gemini 3 Pro (
thinking_level="HIGH") to analyze your incorrect answers. It distinguishes between a "silly mistake," a "conceptual misunderstanding," or a "knowledge gap." - Visible Reasoning: You don't just get the answer; you see the AI's "thought process" as it breaks down your logic in real-time.
- Dynamic Visualizations: If you struggle with a concept, the system writes and executes Python code to generate custom graphs, charts, and data visualizations on the fly.
✨ Key Features
🤖 The Cognitive Companion
- Real-time Diagnostic Reasoning: The AI acts as a private tutor, analyzing your specific misconceptions.
- Socratic Method: Instead of spoon-feeding answers, it asks guiding questions to help you derive the solution yourself.
- Thought Signatures: Maintains context across your session to track your learning progress.
📝 Content-to-Exam Engine
- Generate from Anything: Paste a documentation URL, upload a PDF textbook, or simply type a topic (e.g., "Thermodynamics").
- Structured Output: Automatically generates rigorous, multiple-choice questions with distractors designed to test specific cognitive levels.
👁️ Visual Solver (Multimodal)
- Image Analysis: Upload a photo of a handwritten math problem, a circuit diagram, or a chemical equation.
- Visual Debugging: The AI uses computer vision to identify errors in your work and Code Execution to plot the correct solution visually.
📚 Grounded Learning
- Search Grounding: All AI explanations are cross-referenced with Google Search to ensure facts are accurate and current.
- Citations: Every concept comes with trusted source links.
🚀 Quick Start
Prerequisites
- Node.js 18+ installed
- A Google Gemini API Key (Must have access to
gemini-3-proorgemini-3-flash)
Installation
-
Clone the repository
git clone https://github.com/WAOmaster/Exam-simulator.git cd Exam-simulator -
Install dependencies
npm install # or yarn install -
Configure Environment Create a
.env.localfile in the root directory:GEMINI_API_KEY=your_actual_api_key_hereGet your free Gemini API key from Google AI Studio.
-
Run the application
npm run devOpen http://localhost:3000 in your browser.
📖 How to Use
1. Generate an Exam
- Click "Generate New Exam".
- Paste a link to a Wikipedia page, technical documentation, or upload your study notes (PDF/DOCX).
- Select your difficulty level and click Generate.
- The AI will parse the content and create a structured question set.
2. Practice Mode (Cognitive Diagnostics)
- Start the exam.
- When you answer incorrectly, the Cognitive Companion panel will slide out.
- Watch the "Thinking..." block to see the diagnostic diagnosis.
- Follow the interactive Learning Plan to master the concept before moving on.
3. Visual Solver
- Navigate to the Visual Solver tab.
- Upload an image of a complex diagram or problem.
- The AI will analyze the visual data and provide a step-by-step breakdown using Python-generated graphs where applicable.
🛠️ Tech Stack & Architecture
- Frontend: Next.js 14 (App Router), React, Tailwind CSS, Framer Motion
- AI Orchestration:
- Models:
gemini-3-pro(Reasoning/Diagnostics),gemini-3-flash(Content Generation) - Tools: Code Execution (Python), Google Search Grounding
- Capabilities: Thinking Config (High), JSON Mode, Function Calling
- Models:
- State Management: Zustand with LocalStorage persistence
- Document Processing:
mammoth.js(DOCX),pdf-parse(PDF)
📁 Project Structure
exam-simulator/
├── app/
│ ├── api/ai/diagnose/ # Gemini 3 Deep Think endpoint
│ ├── api/ai/generate/ # Question generation endpoint
│ ├── visual-solver/ # Multimodal vision page
│ └── exam/ # Adaptive exam interface
├── components/
│ ├── CognitiveCompanion.tsx # The diagnostic UI panel
│ ├── CodeVisualizer.tsx # Renders Python-generated graphs
│ └── ThinkingBubble.tsx # Visualizes AI reasoning tokens
├── lib/
│ └── gemini-client.ts # Google AI SDK configuration
└── public/
└── demo-questions.json # Sample question set
📅 Development Timeline
| Phase | Period | Highlights |
|---|---|---|
| v1 - Foundation | Oct 2025 | Core exam engine, question generation from files/URLs/search, AI explanations, practice & exam modes |
| v2 - Polish | Nov 2025 | Theme system (dark/light), export (JSON/CSV/PDF), Learn with AI, question editing, progress tracking |
| v3 - Advanced Input | Dec 2025 | Bulk JSON upload, HOTSPOT & drag-and-drop questions, security patches |
| v4 - Gemini 3 Hackathon | Feb 2026 | Cognitive Companion with deep-think diagnostics, Socratic AI, Visual Solver, AI Study Guide, Gemini 3 integration |
📄 License
This project is open source and available under the MIT License.