zapier
Pass
Audited by Gen Agent Trust Hub on Oct 5, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses a dedicated
zapierCLI binary to perform actions as defined in themanifest.yaml. This binary is used for status checks, authorization, and tool execution. - [INDIRECT_PROMPT_INJECTION]: The skill ingests data from external Zapier tools and schemas, which could contain instructions that influence the agent's behavior.
- Ingestion points: Data retrieved via the
zapier list-toolsandzapier call-toolcommands, which fetch live provider catalogs and tool schemas from Zapier's servers. - Boundary markers: The instructions state that only "reviewed" Zapier agentic-mode tools are exposed, providing a manual curation step before data enters the agent's context.
- Capability inventory: The skill allows the agent to execute read and write methods through the CLI, including managing connections and executing workflows, depending on user approval.
- Sanitization: No specific delimiters or prompt-level sanitization instructions are provided in the skill markdown to handle data returned from the CLI commands.
Audit Metadata