The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill documents mechanisms for ingesting untrusted data from external sources, which presents an attack surface for indirect prompt injection if this data is later processed by an LLM without proper sanitization.
Ingestion points: External data enters the system through push notification payloads (event.data?.json()) and API responses fetched during background sync or periodic sync events, as shown in references/push-and-sync.md.
Boundary markers: The skill does not provide instructions for using boundary markers or safety delimiters when processing or displaying data retrieved from external notifications or background tasks.
Capability inventory: The skill demonstrates capabilities for network communication via fetch, local persistence via the Cache API and IndexedDB, and user interaction through showNotification and openWindow.
Sanitization: The provided code examples for handling push events and background sync do not include sanitization or validation logic for the incoming data payloads.

service-worker