workflows-ai-by-zapier
Pass
Audited by Gen Agent Trust Hub on Oct 16, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill provides guidance for using the
zapier-sdkCLI to discover integration capabilities. These operations are standard for the Zapier platform and facilitate correct tool configuration. - [SAFE]: The instructions prioritize security by directing the agent to use managed credentials and public connection identifiers, explicitly advising against requesting raw user credentials.
- [INDIRECT_PROMPT_INJECTION]: This skill describes a surface for processing untrusted data through the
inputFieldsof an AI step. It includes mandatory safety measures such as usinglockedfields to prevent AI overrides and an explicitfailtool to handle cases where tasks cannot be safely completed. Because these risks are mitigated by design and are central to the skill's legitimate purpose, the verdict is not escalated.
Audit Metadata