directus-flows
Pass
Audited by Gen Agent Trust Hub on Nov 6, 2026
Risk Level: SAFE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill describes mechanisms for ingesting data from external sources such as webhooks and third-party APIs through the 'request' operation. While this creates a surface for indirect prompt injection, the documentation provides explicit security guidance, advising developers to treat '$trigger.body' as untrusted and to avoid passing unsanitized input into operations using elevated permissions ('$full').
- [DYNAMIC_EXECUTION]: The documentation details the 'Run Script' operation, which allows for the execution of custom JavaScript. Security is addressed by noting that execution occurs within a restricted 'isolated-vm' sandbox, which disables access to the file system, network, and sensitive modules like 'require' or 'fetch'.
- [COMMAND_EXECUTION]: Technical instructions include 'curl' command examples for administrative tasks via the Directus REST API. These examples use standard security practices, such as referencing environment variables for authorization tokens and target URLs, rather than hardcoding sensitive information.
- [EXTERNAL_DOWNLOADS]: Mentions official Marketplace operations like '@directus-labs/liquidjs-operation', which are legitimate vendor-provided extensions part of the platform's ecosystem.
Audit Metadata