spark-recipe-new-sender-review
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from external emails using the
spark threadcommand, which could contain malicious instructions intended to influence the agent's behavior. - Ingestion points: Email content is read in
SKILL.mdusing thespark thread <id>command. - Boundary markers: There are no explicit instructions or delimiters provided to the agent to distinguish between email content and system instructions.
- Capability inventory: The skill possesses the capability to take actions based on email content, such as
spark contact-action acceptContact/blockContactinSKILL.md. - Sanitization: No sanitization or validation of the email content is mentioned or implemented within the skill instructions.
Audit Metadata