icon-set-audit
Pass
Audited by Gen Agent Trust Hub on Sep 30, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external SVG files provided by the user, creating a surface where malicious instructions could theoretically be embedded in file metadata or content.
- Ingestion points: The
load_iconfunction inscripts/audit_icons.pyreads and parses SVG files usingxml.etree.ElementTree. - Boundary markers: No explicit boundary markers or instructions to ignore embedded content are used during the data ingestion phase.
- Capability inventory: The script performs file system read operations for the audit and has file system write capabilities when the
--fixargument is provided to modify SVG files. - Sanitization: The script uses standard XML parsing and regular expressions to extract and sanitize numeric data and attributes, which provides some protection against non-numeric injection in coordinate fields.
Audit Metadata