css-border-gradient

Pass

Audited by Gen Agent Trust Hub on Aug 16, 2026

Risk Level: SAFE
Full Analysis
  • [OBFUSCATION]: The demo/index.html file includes a large Base64-encoded string (encodedHtml) that contains the full source for the design demo. This content is decoded at runtime to populate a sandboxed iframe.
  • [EXTERNAL_DOWNLOADS]: The demo environment fetches runtime libraries (Tailwind CSS, GSAP, ScrollTrigger, and Iconify) from well-known services such as cdn.tailwindcss.com, cdnjs.cloudflare.com, and code.iconify.design.
  • [EXTERNAL_DOWNLOADS]: Product and background images for the demo are retrieved from Supabase (hoirqrkdgbmvpwutwuwj.supabase.co), which is a recognized cloud service provider.
  • [DYNAMIC_EXECUTION]: The skill uses srcdoc to render dynamically decoded HTML and employs window.postMessage for injecting asset URLs into the sandboxed iframe. The bundled code uses atob for decoding content.
  • [INDIRECT_PROMPT_INJECTION]: The skill provides an attack surface for indirect prompt injection as it instructions the agent to ingest external design references and product briefs from demo/PROMPT.md to generate UI code.
  • Ingestion points: demo/PROMPT.md (reference URLs and design brief content).
  • Boundary markers: Not present.
  • Capability inventory: The skill provides CSS/HTML templates; no subprocess, network write, or file system write capabilities are present in the skill code.
  • Sanitization: No sanitization or filtering of external design content is performed by the skill instructions.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 16, 2026, 07:33 AM
Security Audit — agent-trust-hub — css-border-gradient