relevancekit
Pass
Audited by Gen Agent Trust Hub on Sep 11, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill provides implementation patterns for watchOS widgets that ingest external data (such as meeting details and task items) to provide contextual relevance clues. This creates a potential attack surface where malicious data could be processed and displayed by the agent or user interface. \n
- Ingestion points: The skill demonstrates data ingestion through methods like
MeetingStore.shared.upcomingMeetings()andTaskStore.shared.dueSoonTasks()inreferences/relevancekit-patterns.md. \n - Boundary markers: The provided code snippets do not include explicit delimiters or instructions to help the system distinguish between code-provided structure and untrusted external content. \n
- Capability inventory: The skill primarily uses standard WidgetKit, MapKit, and HealthKit APIs for UI rendering and relevance signalling; it does not include dangerous capabilities like arbitrary command execution, direct file writes, or network exfiltration within the provided scripts. \n
- Sanitization: The examples do not provide or recommend sanitization, validation, or escaping logic for external strings before they are rendered in the widget views.
Audit Metadata