refactoring-ui

Pass

Audited by Gen Agent Trust Hub on Apr 6, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [SAFE]: Comprehensive analysis of the instructions and associated files confirms the skill is dedicated to educational UI design principles and lacks any malicious patterns, obfuscation, or unauthorized data access mechanisms.
  • [COMMAND_EXECUTION]: The skill includes a benign local utility script, scripts/audit_css.py, used to automate design reviews for the user.
  • Evidence: The script utilizes standard Python libraries (re, pathlib) to perform read-only static analysis on CSS and HTML files.
  • Context: Operations are strictly limited to identifying design token violations and do not involve network exfiltration, arbitrary command execution, or sensitive file access.
  • [PROMPT_INJECTION]: The skill's 'Design Review' mode processes user-provided code, which represents a potential surface for indirect prompt injection (Category 8).
  • Ingestion points: UI code and designs provided by the user in SKILL.md (Design Review mode).
  • Boundary markers: Absent in the provided instructional text.
  • Capability inventory: Limited to read-only file system access via the provided Python audit script.
  • Sanitization: No explicit sanitization or filtering of user-provided design code is performed.
  • Risk Assessment: The risk is negligible as the skill lacks high-risk capabilities like network access or persistent file modification.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 6, 2026, 02:43 AM