mcp-best-practices

Pass

Audited by Gen Agent Trust Hub on Aug 3, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill is a comprehensive instructional guide focused on best practices and security auditing for MCP servers. It contains no hidden or malicious code.
  • [EXTERNAL_DOWNLOADS]: The skill instructs the user or agent to use npx -y @modelcontextprotocol/inspector to audit server metadata. This is the official diagnostic tool for the Model Context Protocol and is a standard industry utility for this purpose.
  • [PROMPT_INJECTION]: The skill includes extensive analysis of prompt injection risks (Indirect Prompt Injection), but specifically from a defensive standpoint. It teaches how to verify that server boundaries are protected against malicious model output.
  • [COMMAND_EXECUTION]: The skill provides transparent bash command examples for using the MCP Inspector. These commands are intended for server diagnostics and do not involve obfuscation or unauthorized system access.
  • [DATA_EXFILTRATION]: No exfiltration patterns were detected. The skill provides guidance on preventing data exposure, such as through audience validation and SSRF guards in server implementations.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 3, 2026, 08:14 AM
Security Audit — agent-trust-hub — mcp-best-practices