vercel-plugin-eval

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.90). The skill instructs installing a third-party plugin via "npx add-plugin https://github.com/vercel-labs/vercel-plugin" and then monitors/greps Claude debug logs (~/.claude/debug/*.txt) to parse "provided additionalContext"/"skillInjection" entries, meaning it fetches and consumes untrusted public GitHub/plugin output and log content that can drive injected skills and agent actions.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.90). The Quick Start uses "npx add-plugin https://github.com/vercel-labs/vercel-plugin -s project -y", which fetches and runs code from that GitHub repository at runtime and the plugin can directly control agent behavior via skill injections, so this external URL is a required runtime-executed dependency.