Landing Page Builder
Pass
Audited by Gen Agent Trust Hub on Mar 3, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: Behavioral instructions define a professional persona and do not include bypass markers or requests to ignore safety filters.
- [DATA_EXFILTRATION]: No sensitive file access or network exfiltration to untrusted domains was found; the skill uses provided context variables for operation.
- [REMOTE_CODE_EXECUTION]: The skill does not download or execute remote scripts and relies on referenced internal tools and APIs.
- [COMMAND_EXECUTION]: No dangerous shell commands or privilege escalation attempts are present in the skill definition.
- [SAFE]: While the skill ingests untrusted user data for landing page generation (e.g., product_description, existing_brief), the risk of indirect injection is mitigated by mandatory approval gates at every step, requiring human sign-off before outputting final artifacts. Capability inventory includes web search and AI generation APIs.
Audit Metadata