report-to-google-slides

Pass

Audited by Gen Agent Trust Hub on Jul 15, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • Local File System Access: The skill is designed to process local HTML files and write outputs to a specified directory (typically /tmp/report_to_google_slides). This behavior is consistent with its role as a report conversion tool.
  • Subprocess Execution: The SKILL.md instructions guide the agent to execute a bundled Python helper script. While this involves command-line execution, the scope is limited to the skill's own internal script and intended functionality.
  • External Dependencies: The Python helper script relies on standard libraries such as beautifulsoup4, pillow, and python-pptx. These are well-known, industry-standard packages used for parsing HTML and generating presentation files.
  • SVG Sanitization: The script implements a robust sanitization process for SVG markup and CSS before rendering charts. This includes blocking potentially risky tags (e.g., <script>, <iframe>) and removing event handlers (on*), which is a good security practice for handling potentially untrusted HTML content.
  • No Remote Operations: The analysis confirmed that the skill does not perform direct network requests or download code from untrusted remote servers; it requires the user to provide a local report file.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 15, 2026, 02:40 AM
Security Audit — agent-trust-hub — report-to-google-slides