Kurral - AI Agent Security Platform

Security Testing for AI Agents

Find vulnerabilities in your MCP agents before attackers do. Automated security assessments with detailed reports.

Start Free View Docs

🔒

Security Scanning

Automated vulnerability detection for SQL injection, path traversal, prompt injection, and more.

📊

Security Score

Track your security posture with a 0-100 score. Monitor improvements over time.

🔄

Session Replay

Compare agent executions side-by-side. Catch regressions before they reach production.

Get started in minutes

# Install the CLI

$ pip install kurral

# Run a security scan

$ kurral scan --agent-url http://localhost:3000

# View results

✓ Found 3 vulnerabilities
  • CRITICAL: SQL Injection in search_orders
  • HIGH: Path Traversal in read_file
  • MEDIUM: Missing Auth in process_refund

Security Score: 45/100