spark-recipe-priority-tuning
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes the
sparkCLI tool to fetch email lists and perform actions on contacts, such as marking them as primary or important. These commands are consistent with the skill's stated purpose of email triage.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes data from an external source (the user's email inbox), which could contain malicious instructions designed to manipulate the agent's behavior.\n - Ingestion points: Steps 1 and 6 involve reading from the
Inboxvia thespark emailscommand.\n - Boundary markers: The skill does not employ delimiters or specific instructions to help the agent distinguish between its system instructions and the content of the emails it reads.\n
- Capability inventory: The agent has the ability to modify contact metadata, such as priority and importance levels, via
spark contact-action.\n - Sanitization: There is no evidence of content sanitization or validation for the data retrieved from the email inbox.
Audit Metadata