kurral 0.4.1

Security testing and observability for AI agents - find vulnerabilities before production

Kurral Security Platform

AI Agent Security Assessment Platform - Proactively find vulnerabilities in your AI agents before attackers do.

Architecture

┌─────────────────────────────────────────────────────────────┐
│                    Kurral Platform                          │
├─────────────────┬─────────────────┬─────────────────────────┤
│   Frontend      │    Backend      │    CLI (kurral)         │
│   (Next.js)     │    (FastAPI)    │    (Python)             │
│   app.kurral.com│    Render       │    pip install kurral   │
└─────────────────┴─────────────────┴─────────────────────────┘

Quick Start

CLI Installation

pip install kurral

Run Security Assessment

# Assess an MCP agent
kurral assess --target http://localhost:3000/mcp

# View results in dashboard
# → app.kurral.com/dashboard

Project Structure

├── frontend/          # Next.js dashboard (Vercel)
├── backend/           # FastAPI API server (Render)
├── kurral/            # Core library (MCP proxy, assessment)
├── kurral_security/   # CLI and assessment tools
├── examples/          # Demo agents (ShopBot)
└── docs/              # Documentation

Development

Frontend

cd frontend
npm install
npm run dev
# → http://localhost:3000

Backend

cd backend
poetry install
uvicorn app.main:app --reload
# → http://localhost:8000

Environment Variables

See:

• frontend/.env.local.example
• backend/.env.example

Deployment

• Frontend: Vercel (auto-deploy from main)
• Backend: Render (Docker, auto-deploy from main)
• Database: Neon PostgreSQL

License

Proprietary - Kurral Security Inc.