The Agent Skills Directory

[SAFE]: The skill contains legitimate development instructions for the iOS App Intents framework. It explicitly guides developers to implement security controls such as authenticationPolicy and requestConfirmation for sensitive actions, adhering to framework-level security best practices.\n- [INDIRECT_PROMPT_INJECTION]: The skill describes an interface for ingesting untrusted data through intent parameters, which is a standard feature of the framework.\n
Ingestion points: Data entering the app via @Parameter properties in AppIntent and AppEntity declarations (e.g., in references/parameters-and-entity-queries.md).\n
Boundary markers: The documentation includes instructions for using requestConfirmation() and IntentAuthenticationPolicy (e.g., .requiresLocalDeviceAuthentication) to guard sensitive logic.\n
Capability inventory: Implementation examples show actions interacting with application services such as OrderService, LightService, and PhotoViewer.\n
Sanitization: Input validation is managed by Swift's strong typing (e.g., UUID, URL), but there is no specific guidance on sanitizing string inputs for downstream AI processing.

app-intents