How is your cognitive state? An open-source IRT-based test that measures AI's impact on your thinking abilities.
Cognitive Rustproof is an open-source cognitive baseline test that helps you regularly measure your core mental abilities — logic reasoning, mental math, and vocabulary.
It uses Item Response Theory (IRT), the same framework behind the SAT and GRE, to adapt question difficulty to your ability level in real time. Each test takes about 15 minutes. Results give you a degradation index (0–100, lower is better).
The real value comes from retesting: take a test every 7+ days and connect the dots. Trends matter more than any single score.
# Install dependencies
pnpm install
# Start development server
pnpm devVisit http://localhost:3000 to preview.
The test includes AI-generated questions with adaptive difficulty. Configure your LLM API:
cp .env.example .env.localEdit .env.local with your API key (any OpenAI-compatible provider — DeepSeek recommended):
| Variable | Description |
|---|---|
OPENAI_BASE_URL |
API endpoint (default: DeepSeek) |
OPENAI_API_KEY |
Your API key |
OPENAI_MODEL |
Model name (default: deepseek-v4-flash) |
NEXT_PUBLIC_ADAPTIVE_MODE |
Set true to enable AI questions |
| Layer | Choice |
|---|---|
| Framework | Next.js 16 (App Router) |
| UI | React 19 + Tailwind CSS v4 + shadcn/ui |
| Package | pnpm |
| Language | TypeScript |
- Official site: cortex.hydroroll.team
- GitHub: github.com/HsiangNianian/cortex
MIT © Hsiang Nianian