The Agent Skills Directory

[SAFE]: No malicious patterns, obfuscation, or unauthorized network operations were identified. The skill focuses on implementing secure architectural patterns for handling user-uploaded media.
[INDIRECT_PROMPT_INJECTION]: The skill handles untrusted data in the form of images and videos ingested from the device's library or camera. However, this is handled securely through mandatory re-encoding which sanitizes the input.
Ingestion points: Media is ingested via MediaPickerPort in templates/src/services/media-picker/media-picker-port.ts and FileTransferPort downloads in examples/src/services/file-transfer/file-transfer-port.ts.
Boundary markers: The system uses a specialized MediaRef type to isolate filenames from URIs, preventing directory traversal or confusion between local and remote sources.
Capability inventory: The skill uses expo-file-system for local storage and FileTransferPort for network uploads/downloads, with capabilities restricted to specific service adapters.
Sanitization: The skill mandates re-encoding picked images using expo-image-manipulator (seen in examples/src/services/image-resizer/expo-image-manipulator-image-resizer-adapter.ts), which strips sensitive EXIF metadata such as GPS coordinates before storage or sharing.

media-images-and-files