sending-transactional-emails
Pass
Audited by Gen Agent Trust Hub on Nov 6, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCREDENTIALS_UNSAFE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to ingest and analyze external data from websites, user-provided HTML, and screenshots to extract branding assets (logos, colors, fonts) and recreate email templates. This creates a surface where external content could potentially influence agent behavior or template generation.
- Ingestion points: Branding setup section in SKILL.md and template migration section in references/moving-from-other-providers.md.
- Boundary markers: No explicit delimiters or instructions to ignore embedded commands are specified for the branding analysis phase.
- Capability inventory: integrations-email-create, workflows-make-email-template-transactional-execute, and media-images-list.
- Sanitization: No explicit sanitization or filtering is mentioned for the branding metadata extraction process.
- [CREDENTIALS_UNSAFE]: The skill guides users through the use of project secret API keys for transactional email sending. It correctly identifies the risk and explicitly instructs users to store keys in environment variables and never to paste them into the chat or client-side code.
Audit Metadata