imagegen-frontend-mobile
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface due to processing untrusted data combined with file modification capabilities.
- Ingestion points: User-supplied design requests enter the agent context in
SKILL.mdto dictate the image-generation parameters and app categories. - Boundary markers: Absent; there are no specific boundary markers or instructions to isolate user-supplied input from the system's operational instructions.
- Capability inventory: In
SKILL.md, the skill mandates file-write capabilities, instructing the agent to create and modify markdown files in thellmwiki/wiki/directory to document execution details. - Sanitization: Absent; the process lacks validation or escaping mechanisms to prevent user-controlled text from corrupting or injecting formatting into the generated wiki files.
Audit Metadata