If you discover a security vulnerability in AGNI, please report it responsibly:

1. Do NOT open a public GitHub issue for security vulnerabilities.
2. Email security@agni-project.org (or open a private security advisory on GitHub).
3. Include:
   • A description of the vulnerability
   • Steps to reproduce
   • Potential impact
   • Suggested fix (if any)

We will acknowledge your report within 48 hours and aim to provide a fix or mitigation within 7 days for critical issues.

AGNI is designed for offline-first village hub deployments on local networks. The threat model assumes:

• The hub runs on a private LAN (e.g., school Wi-Fi or Raspberry Pi hotspot).
• Physical access to the hub device implies trust.
• Internet connectivity is intermittent or absent.

However, we still follow security best practices:

• Creator passwords are hashed with scrypt (64-byte key, random salt).
• Session tokens are cryptographically random (32 bytes).
• Lesson integrity is verified via Ed25519 signatures.
• Sync packages re-pseudonymize student identifiers before transmission.
• Input validation is enforced on lesson content via JSON Schema (Ajv).

• Student PINs are short numeric codes designed for shared-device convenience, not high-security authentication.
• CORS is configured permissively (*) for local network compatibility.
• The hub is not designed for direct internet exposure without a reverse proxy.