icon-set-extend
Pass
Audited by Gen Agent Trust Hub on Oct 2, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses bundled Python scripts (
scripts/infer_spec.pyandscripts/validate_icons.py) to automate the analysis of icon sets and validate new icons. These are standard shell commands executed within the workflow.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes external SVG files, which is an inherent attack surface for indirect prompt injection.\n - Ingestion points: Bundled scripts parse all
.svgfiles in the target directory using thexml.etree.ElementTreelibrary.\n - Boundary markers: There are no explicit delimiters or "ignore" instructions used when the agent interprets the SVG content or the script outputs.\n
- Capability inventory: The agent has permissions to write files (creating new SVGs) and run shell commands (executing the analysis scripts).\n
- Sanitization: The scripts use structural XML parsing to extract geometric data, which provides a layer of isolation from raw text, although they do not specifically filter for natural language instructions hidden in comments or metadata.
Audit Metadata