skills/vercel-labs/emulate/vercel/Gen Agent Trust Hub

vercel

Pass

Audited by Gen Agent Trust Hub on Sep 15, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
  • Command Execution: The skill provides various curl commands and instructions to use npx for starting a local emulator service. These tools are used to interact with a local development server (localhost:4000) and are standard for the skill's purpose of API testing and service emulation.
  • External Package Management: The skill references npx emulate and the @vercel/blob SDK, which involves fetching packages from a public registry. As these are core components for creating a local Vercel-like environment, they are considered routine and expected operations for development tasks.
  • Indirect Prompt Injection Surface: The skill provides examples where user inputs are used to construct API requests and command arguments. While this is necessary for an API tool, it represents a surface where input handling should be managed to ensure commands are processed correctly.
  • Ingestion points: User-provided values for authentication tokens, project settings, and API payloads.
  • Boundary markers: None identified in the skill content.
  • Capability inventory: Bash tool for curl and npx execution.
  • Sanitization: Not explicitly detailed, which is typical for developer-focused emulation tools.
  • Local Test Credentials: The documentation includes placeholder tokens and secrets (e.g., test_token_admin) for use with the local emulator. These are identifiers for development use in an isolated environment and do not involve sensitive production credentials.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 15, 2026, 10:00 AM
Security Audit — agent-trust-hub — vercel