visual-skills

Pass

Audited by Gen Agent Trust Hub on May 20, 2026

Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it processes untrusted structured data from external files without sanitization or boundary markers.
  • Ingestion points: The skill instructions specify reading data from /mnt/user-data/uploads/ via pd.read_csv and pd.read_excel, and also support parsing pasted inline data.
  • Boundary markers: No delimiters or instructions are provided to the agent to treat the data as inert or to ignore embedded instructions.
  • Capability inventory: The skill has the capability to generate and render HTML/JavaScript artifacts directly in the chat interface.
  • Sanitization: There is no requirement or logic provided for escaping or sanitizing data values before they are interpolated into the const DATA object within the HTML <script> tags, potentially allowing malicious data to execute JavaScript (XSS) in the context of the visualization panel.
  • [EXTERNAL_DOWNLOADS]: The visualization templates reference the ECharts library from Cloudflare's official CDN (cdnjs.cloudflare.com). This is a well-known and trusted service used as intended for rendering the charts and does not contribute to verdict escalation.
Audit Metadata
Risk Level
SAFE
Analyzed
May 20, 2026, 06:07 AM
Security Audit — agent-trust-hub — visual-skills