phishing-email-html
Pass
Audited by Gen Agent Trust Hub on Sep 4, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data to generate structured email templates.
- Ingestion points: The workflow relies on a user-supplied 'pretext' (provided via instructions in
SKILL.md) which may contain malicious instructions designed to override the agent's behavior. - Boundary markers: No delimiters or specific formatting requirements are specified for the input data to separate it from the system instructions.
- Capability inventory: The skill is limited to text generation and formatting for delivery platforms (GoPhish, Phishmonger); it lacks file system or network execution capabilities.
- Sanitization: The instructions do not prescribe sanitization or validation of the input pretext content before the agent extracts sender personas and body copy.
Audit Metadata