brand-design-md

Warn

Audited by Gen Agent Trust Hub on Jun 13, 2026

Risk Level: MEDIUMEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill instructs the user or agent to download and install code from an external GitHub repository (zephyrwang6/brand-design-md). This source is not verified or part of the trusted vendor list.
  • [REMOTE_CODE_EXECUTION]: The installation instructions utilize npx skills add, which downloads and executes skill content from a remote repository. Installing unvetted code from third-party developers can lead to arbitrary code execution.
  • [COMMAND_EXECUTION]: The skill provides specific shell commands for environment checking and installation. It highlights the use of the !command prefix (Dynamic Context Injection) to execute these commands directly in the agent's context.
  • [PROMPT_INJECTION]: The skill exposes an attack surface for indirect prompt injection by processing external design tokens from a third-party repository and applying them to project files.
  • Ingestion points: External design tokens fetched from github.com/zephyrwang6/brand-design-md (as described in the functional overview).
  • Boundary markers: Not present.
  • Capability inventory: Capability to modify project CSS variables and Tailwind configuration files.
  • Sanitization: No evidence of validation or sanitization of the remote data before application.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Jun 13, 2026, 10:43 AM
Security Audit — agent-trust-hub — brand-design-md