agent-starter-kit

Warn

Audited by Gen Agent Trust Hub on Sep 21, 2026

Risk Level: MEDIUMEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONREMOTE_CODE_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill's primary functionality is delivered through a bash script (install.sh) that executes multiple system commands, including npx and the claude CLI, to modify the agent's environment.
  • [EXTERNAL_DOWNLOADS]: The installer script fetches and installs external components from multiple GitHub repositories. While some sources are from the vendor or established organizations like Vercel, others are from unverified third-party GitHub accounts.
  • Fetches configuration and skills from Vercel Labs' repositories.
  • Downloads a design-related skill from the vendor's GitHub repository.
  • Installs additional skills and plugins from individual GitHub users (forrestchang, nextlevelbuilder, juliusbrussee, and obra).
  • [REMOTE_CODE_EXECUTION]: The skill uses npx -y skills add and claude plugin install to automatically download and integrate remote code into the agent's execution path. The use of the -y flag in npx bypasses user confirmation for the installation of the skills-management tool itself, and the script proceeds to install remote skills and plugins without version pinning or integrity verification.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Sep 21, 2026, 06:35 PM