split-layout-technical

Pass

Audited by Gen Agent Trust Hub on Aug 26, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [OBFUSCATION]: The file demo/index.html contains a large Base64-encoded string (encodedHtml) that stores the full source code of the design demo, which is decoded and rendered at runtime.
  • [COMMAND_EXECUTION]: The demo loader in demo/index.html dynamically injects the decoded HTML into an iframe's srcdoc attribute, effectively executing bundled code. This is mitigated by the sandbox attribute and a strict Content Security Policy (CSP).
  • [EXTERNAL_DOWNLOADS]: The demo environment is configured to fetch assets from Supabase and runtime scripts from well-known content delivery networks, including Cloudflare (GSAP), Tailwind Labs (Tailwind CSS), and Iconify.
  • [PROMPT_INJECTION]: The demo/PROMPT.md file defines an ingestion surface for indirect prompt injection by instructing the agent to fetch and reference external content from neuform.ai to guide the design process.
  • Ingestion points: demo/PROMPT.md references external design URLs.
  • Boundary markers: None present in the instructions.
  • Capability inventory: The execution environment utilizes a sandboxed iframe and a restricted CSP to limit the impact of untrusted content.
  • Sanitization: No active sanitization is performed on the decoded demo HTML before it is loaded into the design preview.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 26, 2026, 01:37 PM
Security Audit — agent-trust-hub — split-layout-technical