我開發了MCP Guard，因為AI代理直接存取資料庫讓我感到恐懼
Hacker News·
作者開發了MCP Guard，一個託管的SaaS儀表板，旨在提供AI代理行為的可見性和控制力，特別是為了在危險指令執行前加以阻止，這源於對AI代理直接存取資料庫的擔憂。
I built MCP Guard (https://mcp-shield.vercel.app) because I wanted visibility into what my AI agents were actually doing, and control to block dangerous commands before they execute.
It's a hosted SaaS dashboard - no NPM packages, no local proxies to maintain. Just configure security rules through a browser and point your AI client to the secure endpoint.
相關文章