vercel-sandbox

Pass

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [PROMPT_INJECTION] (LOW): The skill is susceptible to Indirect Prompt Injection (Category 8) because it retrieves and processes external content to update its reference files.\n
  • Ingestion points: Documentation is fetched from https://vercel.com/docs/vercel-sandbox and its sub-pages by the update_docs.py script and saved to the references/ directory.\n
  • Boundary markers: The script adds source URL comments to the fetched files, but the skill does not implement robust structural delimiters to prevent the agent from interpreting fetched text as instructions.\n
  • Capability inventory: The skill possesses the ability to create and manage sandboxes, execute commands within them via the Vercel SDK, and perform local configuration tasks.\n
  • Sanitization: The scripts/update_docs.py script employs an HTML parser to extract text and remove potentially active tags such as <script> and <style>.\n- [EXTERNAL_DOWNLOADS] (SAFE): The documentation update mechanism performs network requests to vercel.com. These downloads are legitimate, target a trustworthy primary source for the skill's functionality, and do not involve the execution of remote scripts.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 17, 2026, 06:23 PM