uniapp-project-creator

Fail

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: HIGHEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTION
Full Analysis
  • Remote Code Execution & External Downloads (HIGH): The skill instructs the agent to download project templates from a GitHub repository (dcloudio/uni-preset-vue) which is not on the trusted sources list. These templates are then built or executed using npm install, which is a high-risk behavior for code of unverified origin.
  • Persistence Mechanisms (HIGH): The skill provides explicit instructions to modify user shell profile files (~/.bashrc and ~/.zshrc) to add persistent command aliases. If automated by an agent, this establishes a persistent presence on the user's system.
  • Command Execution & Injection (HIGH): The provided shell scripts (e.g., create-vue2-project.sh in templates/cli-commands.md) are vulnerable to command injection because they interpolate the user-provided $PROJECT_NAME variable directly into shell commands without sanitization.
  • Environment Modification (MEDIUM): The skill encourages high-risk modifications such as global package installations (npm install -g @vue/cli) and changing the global npm registry to an external mirror, which can redirect dependency resolution.
  • Over-privileged Templates (LOW): The project templates provided in templates/project-templates.md include manifest.json files with extensive Android permissions (e.g., READ_LOGS, GET_ACCOUNTS, READ_PHONE_STATE) that may exceed the requirements of a typical application.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Feb 17, 2026, 06:10 PM