html-ppt-zhangzara-bold-poster

Pass

Audited by Gen Agent Trust Hub on Jun 25, 2026

Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [PROMPT_INJECTION]: The skill contains a surface for indirect prompt injection because it is designed to ingest untrusted user data and interpolate it into an HTML-based presentation artifact.
  • Ingestion points: User-provided headlines, body copy, names, and dates are used to replace placeholders within the example.html file.
  • Boundary markers: The instructions do not establish clear delimiters or provide the agent with "ignore embedded instructions" warnings to prevent the interpretation of malicious payloads in the input data.
  • Capability inventory: The skill utilizes the fs:write capability (defined in open-design.json) and the live-artifact mechanism to create and save the resulting HTML file.
  • Sanitization: No specific filtering or validation steps are defined in the skill to sanitize the user input before it is rendered into the final artifact.
  • [EXTERNAL_DOWNLOADS]: The template references external assets from a well-known service.
  • Evidence: The example.html file fetches font files from the Google Fonts API (fonts.googleapis.com).
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 25, 2026, 03:41 AM
Security Audit — agent-trust-hub — html-ppt-zhangzara-bold-poster