email-prompt-building
Pass
Audited by Gen Agent Trust Hub on Oct 26, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill follows an 'architectural principle' of design-time commitment, meaning it generates static markdown templates rather than executing code or performing per-row LLM operations at runtime.
- [SAFE]: No sensitive file paths, hardcoded credentials, or unauthorized network operations were detected. The skill primarily reads local project context files and research data to generate template sets.
- [SAFE]: There are no signs of obfuscation, dynamic code execution, or privilege escalation. The instructions are transparent and focused on marketing copy best practices.
- [SAFE]: The 'Input Fields contract' system uses standard placeholder substitution (e.g.,
{{first_name}}) which is a common and safe pattern for template engines. - [INDIRECT_PROMPT_INJECTION]: The skill processes external data (Company context, Research, Hypotheses). However, it is rated as safe/low risk because it is a design-time tool where the output is a static template for human review, and the architecture specifically avoids 'per-row LLM' execution which is the primary vector for indirect injection attacks. All Cat 8 evidence points to a well-contained design surface.
Audit Metadata