nextjs-og-image

Pass

Audited by Gen Agent Trust Hub on Oct 15, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill provides an architectural pattern for multi-domain support that ingests untrusted HTTP headers (x-forwarded-host and host) to dynamically generate application metadata.
  • Ingestion points: app/layout.tsx ingests headers via the headers() function from next/headers to determine the metadataBase and Open Graph image URLs.
  • Boundary markers: None present; the logic directly interpolates header values into URL construction.
  • Capability inventory: The skill utilizes file system reads (fs.readFile) in assets/og-image-template.tsx to process logo files and executes local build processes (next build).
  • Sanitization: The implementation uses the new URL() constructor, which provides basic URL validation but does not sanitize the origin of the domain provided in the headers.
  • [COMMAND_EXECUTION]: The instructions suggest executing a Python one-liner to extract pixel dimensions from local PNG files using the struct module.
  • Evidence: python3 -c "import struct; d=open('public/images/logo.png','rb').read(); w,h=struct.unpack('>II',d[16:24]); print(f'{w}x{h}')" in SKILL.md.
  • [EXTERNAL_DOWNLOADS]: The documentation includes an example of fetching font data from Google Fonts' official CDN to support custom typography in generated images.
  • Evidence: fetch('https://fonts.gstatic.com/s/inter/v13/UcCO3FwrK3iLTeHuS_fvQtMwCp50KnMw2boKoduKmMEVuLyfAZ9hiJ-Ek-_EeA.woff') in references/satori-gotchas.md.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 15, 2026, 05:42 AM