make-scenario-building
Pass
Audited by Gen Agent Trust Hub on Oct 7, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill facilitates the design of scenarios that ingest data from external sources (such as Webhooks, Google Sheets, and Notion) and process it via AI modules. This architecture creates an attack surface for indirect prompt injection where malicious instructions in the source data could attempt to influence the AI's behavior.
- Ingestion points: Scenarios ingest data through modules like
gateway:CustomWebHook,google-sheets:watchRows, andnotion:watchDatabaseItems. - Boundary markers: The provided blueprint examples (e.g., template 08) use text headers to separate input fields but do not incorporate explicit 'ignore instructions' guardrails in the prompt templates.
- Capability inventory: The skill enables the creation of scenarios that can perform significant actions, including sending emails via Gmail, posting to LinkedIn or Facebook, and updating cloud databases.
- Sanitization: The instructions focus on functional mapping and do not specify techniques for escaping or validating external content for LLM safety.
- [EXTERNAL_DOWNLOADS]: The skill fetches scenario blueprints and reference documentation from Make.com's official public library using platform-native tools. These resources are from a well-known service provider and are used neutrally to provide structural references for building automations.
- [COMMAND_EXECUTION]: The instructions reference the use of the 'Make Code' module (
code:ExecuteCode), which is a standard platform feature for executing custom Javascript logic within a sandboxed environment on the Make.com service. The skill describes the use of this module for legitimate automation tasks.
Audit Metadata