storefront-to-brand-system
Pass
Audited by Gen Agent Trust Hub on Aug 31, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill instructions and reference materials focus entirely on brand design methodology and commercial analysis. There are no attempts to execute system commands, access sensitive files, or perform network operations. The skill uses local Markdown files for its design frameworks and quality checks.
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process user-provided images (storefronts, products) which could theoretically contain adversarial text instructions. However, the risk is strictly limited as the skill does not request or possess high-risk capabilities such as network access or shell command execution.
- Ingestion points: Storefront, interior, and product photos provided by users as spatial anchors (documented in SKILL.md).
- Boundary markers: Not present for the analysis of visual input data.
- Capability inventory: Analysis of space, generation of design concepts, creation of brand assets, and outputting SVG code; no tools for network exfiltration or persistent system modification are present.
- Sanitization: The skill does not implement specific sanitization or filtering for text extracted from processed images.
Audit Metadata