作为第一批90后已经35岁了，虽然早早做了准备，但是好像一直在准备。上次琢磨还是毕业之后靠啥吃饭，现在又到被毕业的阶段了，各种成功手段到我这都成玄学了，因为开始走的就不是人脉路线，就得自己找船，唯一能安慰自己的就是只要眼前的船够多总有一个能上去。

所以就得折腾

1. 文生图

最近Gemini 2.0 flash 不是挺火么，原生文生图很厉害，接口我也测试了，返回的是Base64，需要自己转一下，我用 cloudflare 做了转换和图床，这里记录下别的白嫖平台，智谱 和 硅基流动 还有 image.pollinations.ai 都有免费的文生图服务，生成的效果反正就是抽卡，多抽几次呗；

2. 查询API

智谱的免费查询接口收费了 0.03元每次，博查是 0.036元每次，比 bing 什么的便宜多了，返回的数据还是比较详细的，省着自己再爬，我又部署了firecrawl 效果还不错。别的jina Tavily Brave都有免费额度，但是收费也挺贵的，感兴趣的可以搜一搜。

3. MCP

Cherry Studio支持MCP了，测试了一下，准备自己写服务试试

{
  "mcpServers": {
    "fetch": {
      "isActive": true,
      "description": "fetch爬虫",
      "command": "uvx",
      "args": [
        "mcp-server-fetch"
      ]
    },
    "everything-search": {
      "isActive": true,
      "description": "everything-search",
      "command": "uvx",
      "args": [
        "mcp-server-everything-search"
      ],
      "env": {
        "EVERYTHING_SDK_PATH": "D:\\Soft\\EverythingToolbar-0.5.2\\Everything64.dll"
      }
    },
    "mcp-server-firecrawl": {
      "isActive": true,
      "command": "npx",
      "args": [
        "-y",
        "firecrawl-mcp"
      ],
      "env": {
        "FIRECRAWL_API_URL": "http://148.135.77.70:3002",
        "FIRECRAWL_API_KEY": "fc-search"
      }
    },
    "filesystem": {
      "isActive": false,
      "command": "cmd",
      "description": " Read/write files Create/list/delete directories Move files/directories Search files Get file metadata",
      "args": [
        "/c",
        "bunx",
        "-y",
        "@modelcontextprotocol/server-filesystem",
        "C:\\Program Files\\Cherry Studio"
      ],
      "autoApprove": [
        "read_file",
        "read_multiple_files",
        "list_directory",
        "directory_tree",
        "search_files",
        "get_file_info",
        "list_allowed_directories"
      ]
    }
  }
}