redesign-existing-projects
Pass
Audited by Gen Agent Trust Hub on Sep 6, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to scan and read the user's codebase. This creates a potential surface for indirect prompt injection if the audited code contains instructions intended to manipulate the agent. This risk is inherent to the skill's primary purpose. \n
- Ingestion points: Codebase reading described in the "How This Works" section. \n
- Boundary markers: Absent; no instructions are provided to delimit or ignore instructions found in source files. \n
- Capability inventory: The skill performs file modification and dependency verification. \n
- Sanitization: Absent. \n- [EXTERNAL_DOWNLOADS]: The skill references the well-known
https://picsum.photos/service for fetching placeholder images during the design process, which is considered a safe and standard development practice.
Audit Metadata