Pets
Fail
Audited by Gen Agent Trust Hub on Aug 19, 2026
Risk Level: HIGHPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [PERSISTENCE_MECHANISMS]: The skill's routines documentation in
routines.mdsuggests usingcronjobs to manage recurring reminders for pet care. Modifying system scheduling via cron is a persistence mechanism that can allow instructions or scripts to execute automatically across sessions. - [INDIRECT_PROMPT_INJECTION]: The skill processes and logs user-supplied event descriptions into
log.jsonlfiles and subsequently retrieves this data to generate summaries and visual content, creating a vulnerability to indirect prompt injection. - Ingestion points: User-provided event descriptions and behavior logs captured in
SKILL.mdandbehavior.md. - Boundary markers: None identified; logs are stored without delimiters or protective instructions.
- Capability inventory: The skill has capabilities for file system operations and image generation.
- Sanitization: No sanitization or filtering logic is present for the ingested log data.
Recommendations
- AI detected serious security threats
Audit Metadata