frame-data-chart-nyt

Pass

Audited by Gen Agent Trust Hub on Sep 14, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes user-provided data (CSV, JSON, or descriptive text) and interpolates it into an HTML template for visualization.\n
  • Ingestion points: Ingests raw data inputs defined in SKILL.md and example.md.\n
  • Boundary markers: Lacks explicit delimiters or instructions to treat embedded data as untrusted.\n
  • Capability inventory: Generates interactive HTML and SVG content for video surfaces.\n
  • Sanitization: Does not specify escaping or validation of user-contributed labels or values before rendering.\n- [EXTERNAL_DOWNLOADS]: Loads design assets and styling libraries from well-known services.\n
  • Evidence: example.html includes references to Tailwind CSS CDN and Google Fonts.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 14, 2026, 05:59 PM
Security Audit — agent-trust-hub — frame-data-chart-nyt