redesign-existing-projects
Pass
Audited by Gen Agent Trust Hub on Oct 11, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill requires the agent to ingest and analyze external codebase content, which presents a surface for indirect prompt injection attacks where malicious instructions hidden in the project files could influence agent behavior.
- Ingestion points: The "Scan" step in
SKILL.mdrequires reading the entire codebase, including frameworks and styling methods. - Boundary markers: The instructions lack specific guidance for the agent to use delimiters or to ignore potential instructions embedded within the project files being audited.
- Capability inventory: The skill authorizes the agent to modify the codebase ("Fix" step), providing a write capability that could be targeted by a successful injection.
- Sanitization: There are no validation or filtering steps defined for the data retrieved from the scanned project files.
- [EXTERNAL_DOWNLOADS]: The skill suggests fetching image assets from the
picsum.photosservice for use as placeholders in the design when real assets are not available.
Audit Metadata