skills/vercel-labs/json-render/mcp/Gen Agent Trust Hub

mcp

Pass

Audited by Gen Agent Trust Hub on Sep 15, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • Instructional Processing Surface: The skill defines a tool that allows the AI model to generate UI specifications. This creates an interface for indirect prompt injection, as the model's output is used to render interactive components in the user's environment.
  • Ingestion points: Tool definitions in createMcpApp (SKILL.md).
  • Boundary markers: None explicitly defined in the provided snippets.
  • Capability inventory: The UI component communicates with the host via postMessage and the server utilizes Stdio transports (SKILL.md).
  • Sanitization: Not specified in the current configuration.
  • Command Execution for Server Operations: The configuration examples for environments like Cursor and Claude Desktop utilize npx and tsx to execute the server script. This is a standard procedure for running Node.js-based Model Context Protocol servers locally.
  • Local Resource Access: The server implementation includes logic to read local files (e.g., dist/index.html) using the fs module to serve UI content. This is a common pattern for serving static assets within a development framework.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 15, 2026, 05:32 AM
Security Audit — agent-trust-hub — mcp