When reproducing your code, I noticed that the 1.5M dataset uses a large number of MCP servers that are not available on the official Smithery website. Did you build many of them yourselves, or did you find other MCP server services? Could you give some hints?
When reproducing your code, I noticed that the 1.5M dataset uses a large number of MCP servers that are not available on the official Smithery website. Did you build many of them yourselves, or did you find other MCP server services? Could you give some hints?