Security
UAPK Gateway is designed with security as a core principle, providing defense-in-depth for AI agent operations.
Security Model
Security Principles
1. Zero Trust
- Every request is authenticated and authorized
- No implicit trust based on network location
- Continuous validation of credentials and capabilities
2. Least Privilege
- Agents only get the capabilities they declare
- Capability tokens can further restrict permissions
- Time-limited access by default
3. Defense in Depth
- Multiple layers of security controls
- No single point of failure
- Independent verification at each layer
4. Tamper Evidence
- Cryptographically signed audit logs
- Hash-chained records
- Offline verification capability
Security Topics
-
Threat Model - Understand the threats UAPK Gateway protects against - Threat Model
-
Data Handling - How data is stored, transmitted, and protected - Data Handling
-
Key Management - Managing Ed25519 signing keys securely - Key Management
Security Features
| Feature | Description |
|---|---|
| Ed25519 Signatures | Cryptographic signing of all audit records |
| Hash Chaining | Tamper-evident log chain |
| JWT Capability Tokens | Fine-grained, time-limited permissions |
| Policy Enforcement | Configurable allow/deny/escalate rules |
| Human-in-the-Loop | Approval workflow for sensitive actions |
| Audit Trail | Complete history of all decisions |
Quick Security Checklist
- Use HTTPS in production
- Rotate API keys regularly
- Set short expiry for capability tokens
- Enable approval for sensitive actions
- Verify log chain integrity daily
- Backup signing keys securely
- Monitor for unusual activity
- Review denied actions for patterns
Reporting Security Issues
If you discover a security vulnerability:
- Do not open a public GitHub issue
- Email security@uapk.info with details
- Include steps to reproduce if possible
- We'll respond within 48 hours
Related
- Threat Model - Detailed threat analysis
- Data Handling - Data protection
- Key Management - Key security