ui-color-palette-build-color-system

Pass

Audited by Gen Agent Trust Hub on Jul 27, 2026

Risk Level: SAFE
Full Analysis
  • [PROMPT_INJECTION]: The skill does not contain instructions designed to bypass agent constraints or override safety protocols. All instructions are focused on color palette management and token taxonomy.
  • [DATA_EXFILTRATION]: No sensitive file access (e.g., credentials, SSH keys) or unauthorized network exfiltration patterns were detected. The skill operates on design-related data provided by the user.
  • [REMOTE_CODE_EXECUTION]: There are no patterns involving the download and execution of remote scripts or the use of unsafe evaluation functions.
  • [COMMAND_EXECUTION]: The skill interacts with predefined MCP tools (get_color_system) to process color data. This is a standard and safe use of the platform's tool-calling capabilities.
  • [OBFUSCATION]: The instructions and examples are provided in clear, human-readable markdown. No hidden encoding, zero-width characters, or homoglyphs were found.
  • [DYNAMIC_CONTEXT_INJECTION]: The skill does not use the shell execution syntax (exclamation mark with backticks) in its content.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 27, 2026, 04:22 PM
Security Audit — agent-trust-hub — ui-color-palette-build-color-system