代理之后playwright的MCP不能用了

[guoziyangnb]

我问了cursor，他说

虽然MCP代理连接了，但Playwright工具没有被正确索引，这可能是由于：
序列化错误: 之前看到的TypeError: Object of type ToolAnnotations is not JSON serializable错误
工具发现失败: mcpproxy无法正确发现和索引Playwright的工具
版本兼容性: Playwright MCP版本可能与mcpproxy不兼容

我的电脑是windows系统，nodejs是18.20.6,python版本是3.11的，我在终端输入pip install smart-mcp-proxy是能正常安装成功的，只是最后警告了机器学习的库，比如pytorch,tensorflow要求numpy的版本要小于2.0,这个应该不影响，
我的mcp有关的配置如下供您参考是否有问题
mcp.json
{
"mcpServers": {
"mcp-proxy": {
"command": "mcpproxy",
"env": {
"MCPPROXY_CONFIG_PATH": "C:/Users/lenovo/.cursor/mcp_proxy.json"
}
}
}
}
mcp_proxy.json
{
"mcpServers": {
"playwright": {
"command": "cmd",
"args": [
"/c",
"npx",
"-y",
"@playwright/mcp@latest"
]
},
"context7": {
"command": "cmd",
"args": [
"/c",
"npx",
"-y",
"@upstash/context7-mcp"
]
},
"filesystem": {
"command": "cmd",
"args": [
"/c",
"npx",
"-y",
"@modelcontextprotocol/server-filesystem",
"E:/test"
]
},
"github": {
"command": "cmd",
"args": [
"/c",
"npx",
"-y",
"@modelcontextprotocol/server-github"
],
"env": {
"GITHUB_PERSONAL_ACCESS_TOKEN": "xxxxxxxxxxxxxxxxxxx"
},
"disabled": false,
"autoApprove": []
}
}
}
由于我是windows电脑，所以cmd命令是最适合windows的