designing-email-templates
Pass
Audited by Gen Agent Trust Hub on Oct 27, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill provides instructions for using the
curlcommand-line tool to upload image files. This is a standard procedure within the platform's workflow for handling media assets via presigned URLs returned by internal tools. - [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied content, such as email text and Liquid templating tags, which constitutes an attack surface for indirect prompt injection. However, the instructions emphasize the use of standard Liquid escaping and provide guidance on safely handling variables.
- Ingestion points: User-provided email designs, subject lines, and body text processed in
SKILL.md. - Boundary markers: Instructions specify that content should be passed directly in tool calls using standard JSON string escaping.
- Capability inventory: The skill utilizes tools for template creation (
workflows-create-email-template), modification (workflows-patch-email-template), and image management (media-image-upload-start). - Sanitization: The documentation correctly identifies that Liquid output is HTML-escaped by default and warns against the unsafe use of the
rawfilter on untrusted text.
Audit Metadata