Skip to main content
Glama
YouTube Comments MCP Server

YouTube Comments MCP Server

by 878787ya

yt_mcp — YouTube Comments MCP Server

一個遵循 Model Context Protocol (MCP) 的本機伺服器。提供工具:

  • fetch_comments(videoUrl, order="relevance|time", max=300) 以 YouTube Data API v3 取得並展開 頂層留言+回覆,回傳 JSON 字串。


需求

  • Python 3.10+

  • 已啟用 YouTube Data API v3 的 API 金鑰


Related MCP server: mcp-server-youtube

安裝

uv venv
source .venv/bin/activate
uv pip install "mcp[cli]" httpx python-dotenv

設定金鑰(擇一)

A. 用 .env(推薦)

在專案根目錄建立 .env:

YOUTUBE_API_KEY=YOUR_API_KEY_HERE

B. 寫在客戶端設定的 env(見下方範例)


在 Claude Desktop 使用(本機 STDIO)

  1. 開啟:Settings → Developer → Local MCP servers → Edit Config

  2. 加入(請把絕對路徑換成你的實際路徑):

{
  "mcpServers": {
    "yt_comments": {
      "command": "<abs path>/yt_mcp/.venv/bin/python",
      "args": ["<abs path>/yt_mcp/server.py"],
      "env": { "YOUTUBE_API_KEY": "YOUR_API_KEY" }  // 若已用 .env,可移除此段
    }
  }
}

也可不填 env,改用專案根目錄的 .env。

  1. 回到 Local MCP servers 啟動 yt_comments。

  2. 新開對話請 Claude 呼叫:

yt_comments.fetch_comments(
  videoUrl="https://www.youtube.com/watch?v=XXXXXXXXXXX",
  order="relevance",
  max=300
)

接著請模型做摘要/分類/翻譯並輸出報告。


在 MCP Inspector 測試

  • Transport:STDIO

  • Command:<abs path>/yt_mcp/.venv/bin/python

  • Arguments:<abs path>/yt_mcp/server.py

  • Environment(可選):YOUTUBE_API_KEY=YOUR_API_KEY


輸出格式(節選)

{
  "video_id": "abcdEFGhijk",
  "order": "relevance",
  "requested": 300,
  "total_returned": 278,
  "items": [
    {
      "id": "...",
      "parentId": null,
      "author": "Somebody",
      "publishedAt": "2025-01-01T12:34:56Z",
      "likeCount": 42,
      "text": "Great video!"
    }
  ]
}

Troubleshooting

  • Missing YOUTUBE_API_KEY 在 .env 或設定的 env 補上金鑰,並確認金鑰已啟用 YouTube Data API v3。

  • 找不到 server.py 在設定裡把 args 改用絕對路徑(例如 "/Users/you/yt_mcp/server.py")。

  • ModuleNotFoundError 確認已在虛擬環境內安裝相依套件("mcp[cli]" httpx python-dotenv)。

  • HTTP 403/400 可能是配額不足、影片關閉留言、或金鑰權限未開。


安全

  • .env 已在 .gitignore 中,請勿提交金鑰。

  • 若曾不小心提交金鑰,請 旋轉金鑰 並清理 Git 歷史。


授權

MIT License。

Available Tools

1 tool
fetch_commentsA

Fetch public comments for a YouTube video and return a JSON string. Args: videoUrl: Full YouTube video URL. order: "relevance" (default) or "time". max: Max total comments to return (100–1000 建議).

Inputs: videoUrl (required), order, max. Returns: result

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    TDQS

    A3.9/5.0

    Scored across 1 tool

    Disambiguation5/5

    With only one tool, there is no possibility of confusion or overlap between tools.

    Naming Consistency5/5

    A single tool named 'fetch_comments' follows a clear and consistent verb_noun pattern.

    Tool Count3/5

    One tool is borderline; it is appropriate for a highly focused purpose, but most servers benefit from at least a few related tools.

    Completeness2/5

    The server only provides fetching of comments, lacking common operations like replying, deleting, or retrieving replies, which are expected for a YouTube Comments server.

    Maintenance

    ActivityInactive
    ResponsivenessOutdated data

    Related MCP Connectors

    Related MCP Servers