AI 导读
我同意。对大多数集成来说，MCP 明显比 CLI 更好。早期就有这种感觉，现在依然如此。而且前沿模型在这方面还在不断变强。 如果你构建自定义 harness，就用 MCP 工具。 我的 meta harness 所有外部工具都用 MCP。https://t.co/3kWuovtfJy
正文
I agree. MCP is clearly better than CLI for most integrations. It felt that way early on, and it still does now. And frontier models keep getting better at it.
If you build custom harnesses, use MCP tools.
My meta harnesses use MCP for all external tools. https://t.co/3kWuovtfJy
I was not expecting things to go this way, but I think MCPs are better than CLIs for most integrations. The models have gotten much better at tool calling, we can defer tools & MCP is now stateless. If you need to compose/filter data, add params like query to your MCP tools.在 X 查看被引用的帖子
来源：@omarsar0 · x.com