Skip to content
TheChatArchiveThe Chat Archive
o4-mini-2025-04-16· OpenAIHugging Face LMArena Human Preference 140K

Untitled conversation

From a Chatbot Arena battleThe same person sent this prompt to two anonymous models at once. Compare it with chatgpt-4o-latest-20250326’s answer.
  1. User

    { "mcpServers": { "tavily-mcp": { "command": "npx", "args": [ "-y", "tavily-mcp@0.2.1" ], "env": { "TAVILY_API_KEY": "{ "mcpServers": { "tavily-mcp": { "command": "npx", "args": [ "-y", "tavily-mcp@0.2.1" ], "env": { "TAVILY_API_KEY": "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" }, "disabled": false, "autoApprove": [] } } }" }, "disabled": false, "autoApprove": [] } } }

  2. Assistant

    It looks like your JSON got a bit nested inside the TAVILY_API_KEY string. Here’s a cleaned-up version where the API key is just a simple string value:

    {
      "mcpServers": {
        "tavily-mcp": {
          "command": "npx",
          "args": [
            "-y",
            "tavily-mcp@0.2.1"
          ],
          "env": {
            "TAVILY_API_KEY": "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"
          },
          "disabled": false,
          "autoApprove": []
        }
      }
    }
    

    Make sure:

    1. Your JSON file has no trailing commas.
    2. The TAVILY_API_KEY value is just the raw key string.
    3. You’ve saved and reloaded any processes that read this config.

    Let me know if that fixes it or if you run into any other issues! 😃

  3. User

    这个代码在chatwise里可以使用,但我用在cherry studio里不行,请分析错误,给出正确的配置

  4. Assistant

    下面给出两种常见的做法,一是把完整的 mcpServers 写入 Cherry Studio 的配置文件(JSON 或 YAML),二是直接在 Cherry Studio 的 Web UI 里注入环境变量后,在配置里以占位符引用。你可以根据自己项目实际选一种即可。


    一、直接在配置里写死(JSON)

    保存为 cherry.config.json(或你项目实际读取的文件名):

    {
      "mcpServers": {
        "tavily-mcp": {
          "command": "npx",
          "args": [
            "-y",
            "tavily-mcp@0.2.1"
          ],
          "env": {
            "TAVILY_API_KEY": "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"
          },
          "disabled": false,
          "autoApprove": []
        }
      }
    }
    

    要点:

    1. 不要在字符串里再塞一段 JSON —— 原先的双层大括号把整个对象当成了一个字符串,导致 Cherry Studio 解析失败。
    2. 确保没有多余的逗号或引号,文件整体是一个合法的 JSON。

    二、借助 Cherry Studio 的环境变量功能(推荐)

    1. 在 Cherry Studio 项目设置 → Environment Variables → 添加:
      • Key:TAVILY_API_KEY
      • Value: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
    2. 配置文件里这样写(支持 JSON/YAML 两种形式):

    --- JSON 版(cherry.config.json)

    {
      "mcpServers": {
        "tavily-mcp": {
          "command": "npx",
          "args": ["-y", "tavily-mcp@0.2.1"],
          "env": {
            "TAVILY_API_KEY": "${TAVILY_API_KEY}"
          },
          "disabled": false,
          "autoApprove": []
        }
      }
    }
    

    --- YAML 版(cherry.config.yml)

    mcpServers:
      tavily-mcp:
        command: npx
        args:
          - -y
          - tavily-mcp@0.2.1
        env:
          TAVILY_API_KEY: ${TAVILY_API_KEY}
        disabled: false
        autoApprove: []
    

    这样做的好处:

    • 安全:密钥不会写在代码里,Cherry Studio CI/CD 会自动注入。
    • 灵活:不同环境(dev/stage/prod)可以配置不同的变量。

    按上面任一方案改完后,重启/重新触发你的 Cherry Studio 流水线,就能正常跑起来了。有什么问题继续吼我!😄

Responses were produced in different, uncontrolled contexts (system prompts, settings, prior turns). This is not a benchmark.