Read this in: English · 简体中文
Paste any text. Get an interactive knowledge graph in seconds.
Built for analysts, journalists, and researchers who need to understand complex entity relationships, without spending 20 minutes drawing them by hand.
- Extract: Feed it any text (news articles, intelligence reports, research papers). An LLM extracts all entities and relationships automatically.
- Visualize: Renders an interactive force graph. Drag nodes, zoom, click to inspect relationships.
- Infer: Hit the INFERENCE button for a strategic analysis report: power centers, influence paths, hidden patterns, predictions.
Works with any OpenAI-compatible API: DeepSeek, Qwen, GPT-4o, local models via Ollama, etc.
Backend
cd backend
cp .env.example .env
# Edit .env, fill in your API key and base URL
pip install -r requirements.txt
uvicorn main:app --reload --port 8000Frontend
cd frontend
cp .env.example .env # optional: change VITE_API_URL if backend runs elsewhere
bun run dev # or: npm run dev / npx viteOpen http://localhost:5173
backend/.env
OPENAI_API_KEY=your-key-here
OPENAI_BASE_URL=https://api.deepseek.com/v1 # any OpenAI-compatible endpoint
OPENAI_MODEL=deepseek-reasoner # supports R1-style reasoning chains
ALLOWED_ORIGINS=http://localhost:5173 # comma-separated if multiplefrontend/.env (optional)
VITE_API_URL=http://localhost:8000- DeepSeek R1 / V3
- GPT-4o / GPT-4o-mini
- Qwen 2.5
| Layer | Tech |
|---|---|
| Frontend | React 19 + Vite + D3.js |
| Backend | FastAPI + OpenAI SDK |
| Graph | D3 force simulation |
| Styling | Custom CSS, Share Tech Mono |
MIT