ui-ux-pro-max

Warn

Audited by Gen Agent Trust Hub on Nov 20, 2026

Risk Level: MEDIUMPERSISTENCEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [PERSISTENCE]: The design system generation logic in scripts/design_system.py contains a path traversal vulnerability. The persist_design_system function uses the user-provided project_name or search query to construct a file path for saving the design system. The sanitization logic only replaces spaces with hyphens, which fails to filter path traversal sequences like ../. This could allow a malicious query to write Markdown files into arbitrary locations on the local file system.
  • [INDIRECT_PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it ingests untrusted user input via command-line arguments and writes it directly into Markdown documentation (MASTER.md and page-specific overrides). The skill's primary instructions guide the AI agent to treat these persisted files as a 'source of truth' for future tasks. A malicious payload provided in a query can thus influence the agent's future logic.
  • Ingestion points: Search queries and project names provided to the scripts/search.py CLI (referenced in SKILL.md).
  • Boundary markers: Absent; user-provided data is written directly into the markdown body of generated files.
  • Capability inventory: File system write access (scripts/design_system.py).
  • Sanitization: Absent in the persistence formatting logic.
  • [COMMAND_EXECUTION]: The skill requires the execution of local Python scripts (scripts/search.py and scripts/core.py) to perform its primary search and generation functions. These scripts handle user input for database lookups.
  • [EXTERNAL_DOWNLOADS]: The skill's CSV data files and documentation reference various external resources, including official framework documentation and script CDNs (e.g., cdnjs.cloudflare.com, fonts.googleapis.com). These references point to established, well-known services and are used to provide implementation examples to the user.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Nov 20, 2026, 02:08 AM