vercel-sandbox

Warn

Audited by Snyk on Mar 10, 2026

Risk Level: MEDIUM
Full Analysis

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

  • Third-party content exposure detected (high risk: 0.90). The SKILL.md demonstrates runtime use of agent-browser to open arbitrary URLs (e.g., screenshotUrl/open, snapshotUrl/open, and fillAndSubmitForm which parses page snapshots and fills/clicks elements), so the agent ingests and acts on untrusted public web content that could contain instructions influencing subsequent actions.

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

  • Potentially malicious external URL detected (high risk: 0.90). The skill runs runtime install commands ("npm install -g agent-browser" and "npx agent-browser install") inside the Vercel Sandbox which fetch and execute remote code from the npm registry (https://registry.npmjs.org/) and are required for the skill to operate.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Mar 10, 2026, 08:33 PM