stitch-manage-design-system
Pass
Audited by Gen Agent Trust Hub on Oct 7, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external HTML and design document content, creating a potential surface for indirect injection.\n
- Ingestion points: Content is retrieved from local DESIGN.md files and remote URLs via the read_url_content tool.\n
- Boundary markers: The instructions lack specific delimiters or warnings to ignore embedded instructions within the ingested content.\n
- Capability inventory: The skill has tools to upload, create, and apply design systems to screens.\n
- Sanitization: Detailed instructions are provided to mask customer data (names, phones, orders) with demo values before processing.\n- [EXTERNAL_DOWNLOADS]: The skill references documentation and snapshots from a public GitHub repository under the google-labs-code organization, which is a well-known entity.
Audit Metadata