technical-wireframe-info-layout
Pass
Audited by Gen Agent Trust Hub on Oct 14, 2026
Risk Level: SAFEOBFUSCATIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [OBFUSCATION]: The file
demo/index.htmlcontains a large Base64-encoded string (encodedHtml) representing the complete HTML structure of the interactive design demo. This is a common packaging method used by design vendors to bundle previews into a single loader script. - [EXTERNAL_DOWNLOADS]: The skill's demo configuration in
demo/source.jsonand the decoded demo template reference external assets and scripts from well-known services, including Cloudflare's CDN, Tailwind CSS, Google Fonts, and Fontshare. These are standard resources for modern web development. - [INDIRECT_PROMPT_INJECTION]: The skill's workflow involves generating HTML designs from a combination of external asset metadata and user-supplied prompts. This creates a surface for processing untrusted data, though the risk is addressed through technical controls in the demo environment.
- Ingestion points: Metadata in
demo/source.jsonand user-driven prompts indemo/PROMPT.md. - Boundary markers: No explicit delimiters are used to separate instructional content from untrusted input data.
- Capability inventory: The system generates and executes HTML/JavaScript within a sandboxed
<iframe>managed bydemo/index.html. - Sanitization: The demo environment implements a restrictive Content Security Policy (CSP) via the
srcdocattribute to limit the scope of execution. - [DYNAMIC_EXECUTION]: The demo loader in
demo/index.htmldynamically decodes the Base64 payload and injects it into an iframe using thesrcdocattribute. It also usesURL.createObjectURLto map base64-encoded assets to runtime URLs, which is a standard approach for handling bundled media in sandboxed environments.
Audit Metadata