authentication

name	authentication
description	Enforce standard authentication for external data source connections. Use when integrating, configuring, or reviewing connections from AI systems to external databases, APIs, or data services.
metadata	{"category":"secure_development","subcategory":"external-data-source"}

Authentication for External Data Sources

Security Requirement

External data sources MUST require that the principal (user or service) connecting to them is identified and authenticated. The authentication MUST use standard protocols.

Standard Authentication Protocols

Protocol	Use Case
OAuth 2.1 / OIDC	Web APIs, REST services, cloud data sources
SPIFFE/SPIRE + mTLS	Service-to-service connections within infrastructure
SAML	Federated enterprise data sources
Kerberos	On-premise databases and services (Active Directory environments)
Client certificates (mTLS)	Machine-to-machine data source access

What Is Not Acceptable

• Unauthenticated connections to data sources containing any data
• Shared credentials used by all services
• Credentials embedded in connection strings in source code
• Long-lived API keys without rotation (see api_keys/avoid-api-keys)

Implementation Checklist

• Identify all external data sources accessed by the AI system
• Require authentication for every data source connection
• Use standard authentication protocols (OAuth, mTLS, Kerberos)
• Use unique credentials per service or workload (no shared credentials)
• Store credentials in a secret management system
• Rotate credentials on a regular schedule
• Log authentication events (successes and failures)