alarmkit
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill facilitates the ingestion of untrusted data through alarm and timer metadata fields such as titles and labels.\n
- Ingestion points: Data enters the system via the
AlarmandAlarmAttributesinitializers inSKILL.mdandreferences/alarmkit-patterns.md.\n - Boundary markers: There are no explicit delimiters or instructions provided to the agent to treat these strings as untrusted or to ignore any commands embedded within them.\n
- Capability inventory: The skill utilizes
AlarmManager.shared.scheduleto trigger system-level alerts andAlarmManager.shared.alarmUpdatesto observe and process current alarm states.\n - Sanitization: The implementation patterns do not include sanitization or validation of these strings, which could allow a malicious string to influence the agent when it observes the active alarms list.
Audit Metadata