From hackathon idea to complete strategy in under 3 minutes.
7 AI specialists research, analyse, architect, and document your project
before you write a single line of code.
🚀 Live Demo · Architecture · Specialists · Quick Start
exHacker is an AI-powered hackathon strategy engine built by participants, for participants. Paste any hackathon challenge and watch 7 AI specialists work together to produce:
- 🧠 Challenge Intelligence — Deep analysis of constraints, opportunities, and strategy
- 🔍 Research — 10-category market research with synthesis
- 🎯 Competitor Analysis — Gap analysis, white space, differentiation opportunities
- 💡 Scored Ideas — 5 differentiated concepts with 8-dimension scoring
- 🏗️ Architecture Blueprint — System design with Mermaid diagrams, API contracts, trade-offs
- 📝 Documentation Package — PRD, README, tech stack, API docs, database design, pitch guide
- ⏱ Decision Timeline — Every decision tracked and explainable
Frontend: https://exhacker-frontend.vercel.app/
API: https://exhacker-backend.vercel.app/health
API Docs: https://exhacker-backend.vercel.app/docs
| Feature | Description |
|---|---|
| Live Architecture Evolution | Watch your system design being built component-by-component |
| Project Readiness Score | Real-time 0-100 score with 5 sub-scores in the sidebar |
| Command Palette | Ctrl+K to navigate across all workspace pages |
| Idea Comparison | Side-by-side table comparing 10 scores, features, and effort |
| Mermaid Diagrams | All architecture diagrams rendered as interactive SVGs |
| GitHub-Style Docs Viewer | Rendered markdown with syntax highlighting and inline Mermaid |
Frontend (Next.js 16 + React 19)
├── /app → 12 pages (landing, workspace × 10, projects)
├── /components → 4 domains (pipeline, shared, diagrams, markdown)
└── /services → API client with mock fallback
Backend (FastAPI + Python 3.11)
├── /api/v1 → 10 routers, 40 endpoints
├── /services
│ ├── /specialists → 7 AI specialist implementations
│ └── /shared → Memory, journal, context API
└── /models → 10 SQLAlchemy models
| # | Specialist | Purpose | Model Tier |
|---|---|---|---|
| S1 | Challenge Analyst | Extract themes, constraints, opportunities | Tier 2 (glm-5.2) |
| S2 | Research Specialist | 10-category market research | Tier 1 (deepseek) + Tier 2 |
| S3 | Competitor Analyst | Gap analysis, white space, differentiation | Tier 2 |
| S4 | Shared Intelligence | Memory + decision journal + context API | Tier 0 (deterministic) |
| S5 | Idea Generator | 5 scored ideas with self-critique | Tier 2 × 2 |
| S7 | Solution Architect | Full technical blueprint with Mermaid | Tier 2 |
| S13 | Documentation Writer | 10 deterministic document templates | Tier 0 (templates) |
- Python 3.11+
- Node.js 22+
- SQLite (bundled with Python)
cd backend
python -m venv .venv && source .venv/bin/activate
pip install -e .
# Mock mode (no AI API key needed):
MOCK_AI=true MOCK_RESEARCH=true uvicorn app.main:app --reloadcd frontend
npm install
npm run dev| Variable | Required | Default | Purpose |
|---|---|---|---|
MOCK_AI |
No | false |
Bypass AI APIs with mock responses |
MOCK_RESEARCH |
No | false |
Bypass search APIs with mock data |
NEXT_PUBLIC_API_URL |
No | http://localhost:8000/api/v1 |
Backend URL for frontend |
# Backend
cd backend && pytest -v
# Frontend (type check)
cd frontend && npx tsc --noEmit
# CI (GitHub Actions)
# Triggered automatically on push to main| Layer | Technology |
|---|---|
| Frontend | Next.js 16, React 19, TypeScript, lucide-react |
| Backend | FastAPI, Python 3.11, SQLAlchemy, SQLite |
| AI | LiteLLM gateway, DeepSeek V4, GLM 5.2 |
| Diagrams | mermaid.js, react-markdown, react-syntax-highlighter |
| Infrastructure | Vercel (frontend + serverless API), GitHub Actions |
MIT © Muneer Alam
Built during a hackathon, for hackathon participants. Inspired by the belief that great ideas deserve a great strategy before a single line of code is written.