spark-persona-freelancer
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to read and process external emails, which serves as an ingestion point for untrusted data. A malicious email could contain instructions intended to override the agent's behavior, potentially leading to unauthorized drafting or modification of communications.
- Ingestion points: External email content is ingested through
spark emails,spark search, andspark threadcommands. - Capability inventory: The skill includes write/modification capabilities such as
spark draftfor creating replies andspark action changeReminderfor updating reminders. - Boundary markers: The instructions lack specific delimiters or warnings to the agent to ignore instructions embedded within the processed email bodies.
- Sanitization: There is no explicit requirement for the agent to sanitize or filter email content before interpreting it for next steps.
Audit Metadata