MCP YouTube Transcript
An MCP server retrieving transcripts of YouTube videos
作者提供的说明(英文原文)
Find YouTube outliers, videos that beat their channel's own baseline, on your own free API key.
这三项永远不会合并成一个分数:意见、测量和人气是不同的东西。
我们还没有为它写通俗的介绍。下面是我们从注册中心读取到的事实。
选择你使用的 AI 应用,然后复制设置。
YOUTUBE_API_KEY标为机密的值是你账户的钥匙。千万不要分享给别人,也不要粘贴到聊天里。
你需要
你用的是哪个应用?
在上方选择你的应用。我们会记住它,下次按钮会显示“安装到 …”。
打开 Claude Desktop → Settings → Developer → Edit Config,把下面的内容粘贴到 claude_desktop_config.json
{
"mcpServers": {
"bangermap": {
"command": "npx",
"args": [
"-y",
"bangermap-mcp"
],
"env": {
"YOUTUBE_API_KEY": "your-value"
}
}
}
}在终端中运行这条命令
claude mcp add bangermap --env YOUTUBE_API_KEY=your-value -- npx -y bangermap-mcp运行前请先填写空着的值。
粘贴到项目中的 .cursor/mcp.json,或 ~/.cursor/mcp.json
{
"mcpServers": {
"bangermap": {
"command": "npx",
"args": [
"-y",
"bangermap-mcp"
],
"env": {
"YOUTUBE_API_KEY": "your-value"
}
}
}
}浏览器会先询问是否打开应用,然后应用会显示设置并请你确认。密钥的值会留空,由你自己填写。
粘贴到项目中的 .vscode/mcp.json
{
"servers": {
"bangermap": {
"type": "stdio",
"command": "npx",
"args": [
"-y",
"bangermap-mcp"
],
"env": {
"YOUTUBE_API_KEY": "your-value"
}
}
}
}浏览器会先询问是否打开应用,然后应用会显示设置并请你确认。密钥的值会留空,由你自己填写。
这个服务器在你的电脑上运行,而 ChatGPT 只能连接互联网上的服务器。请选择其他应用。
你的选择通过 Cookie 保存在这台设备上,无需账户。
这些设置是根据作者在注册中心的条目生成的。如果 README 的写法不同,请以 README 为准。
它在你的电脑上运行;安装后,工具列表会显示在你的 AI 应用中。
我们检查的是任何人都能看到的信息;我们没有阅读代码。不适用的检查项不计入总数。 · 检查于 2026年11月7日
注册名称已通过 GitHub 账户 AdesiaHQ 验证,该账户拥有此仓库
我们还没有从 GitHub 读取这个仓库
我们还没有从 GitHub 读取这个仓库
它在你的电脑上运行,没有互联网端点
在你的电脑上运行,改用本地设置
它在你的电脑上运行;我们只能读取互联网端点的工具列表
与注册中心里这个分类中所有活跃的服务器比较。只有当发布者拥有该仓库时,星标数才计入。
它在 1 项适用的基础检查中通过了 1 项。我们没有阅读代码。在添加密钥或令牌之前,请先看看这个服务器要求哪些权限,并使用只具备所需权限的密钥。
你需要填写:YOUTUBE_API_KEY
它在你的电脑上运行。你的 AI 应用会在需要时启动它,数据不会经过作者的服务器(除非它自己调用在线服务)。
GitHub 仓库最后更新于–,最新版本(1.0.3)于3个月前进入注册中心。
支持 MCP 的应用,例如 Claude Desktop、Claude Code、Cursor 和 VS Code。你的电脑上还需要有 Node.js。
同一分类中有:MCP YouTube Transcript, Google Analytics MCP, BibiGPT YouTube Video Summarizer
同一分类中其他作者的服务器,按星标数排序。
如何看卡片:★ GitHub 星标i三大支柱:评价 · 检查 · 使用情况i更新 = 最新版本发布到注册中心的时间i
An MCP server retrieving transcripts of YouTube videos
GA4 MCP server for AI agents: schema discovery, server-side aggregation, analysis-ready defaults.
AI video summarizer: summarize and transcribe YouTube, Bilibili, TikTok and podcast videos.
按原文从作者的仓库复制。PopNx MCP 没有审查或背书其内容。HTML、图片和脚本已被移除。链接会离开本站。
YouTube outlier research over MCP, running on your own free YouTube Data API key.
An outlier is a video that beat its own channel's normal performance. A 10x on a small channel is a stronger signal than a million views on a channel that always gets a million, because it is proof an idea worked without an audience behind it. This server measures that, from any MCP client.
No account, no credits, no meter. The data comes from Google's public API on your key, so the only quota is the free 10,000 units a day Google already gives you, which is roughly 1,100 channel scans.
Create a free key in the Google Cloud console, enable YouTube Data API v3 on its project, then point your client at the server.
Claude Code:
claude mcp add bangermap --env YOUTUBE_API_KEY=your-key -- npx -y bangermap-mcp
Claude Desktop, in claude_desktop_config.json:
{
"mcpServers": { "bangermap": { "command": "npx", "args": ["-y", "bangermap-mcp"], "env": { "YOUTUBE_API_KEY": "your-key" } } } }
Any other MCP client works the same way. The server speaks stdio and needs only YOUTUBE_API_KEY in its environment.
| Tool | What it answers | Cost |
|---|---|---|
find_outliers | Which of this channel's recent uploads beat its own baseline, and by how much | ~5 units |
channel_baseline | What normal looks like for this channel, so a view count can be judged | ~5 units |
compare_channels | What is working across a set of competitors, each measured against itself | ~5 units per channel |
sweep_niche | Which channels a seed channel is connected to, and what is overperforming on them | ~5 units per channel |
Channels can be given as a URL, an @handle, a channel id, or the URL of any video on the channel.
Discovery runs through the channel graph rather than keyword search on purpose. Google caps search at 100 calls a day on a free key while channel reads cost a single unit, so seeding from a channel you already know is both cheaper and deeper.
For each channel, the baseline is the median views of its recent uploads of the same format, counting only uploads old enough to have settled. Shorts and long-form get separate baselines, since comparing a 40-second Short against a 20-minute upload would flatter one and punish the other. A video's multiple is its views divided by that baseline.
The outlier multiple and the baseline are computed by this server, not by YouTube. View and channel data comes from the YouTube Data API. YouTube is a trademark of Google LLC.
This server shares its scoring engine with Bangermap, a Mac app for the same research when you want it to persist. The app keeps a watchlist that re-scans and flags what is new since your last visit, shows outliers in a thumbnail wall for pattern study, exports to CSV, and caches locally so repeat research does not re-spend quota. One payment, no subscription, same free key.
There is also a free browser version at bangermap.com/tools/outlier-finder if you want to try the measurement without installing anything, and an n8n community node, n8n-nodes-bangermap, for the same operations inside a workflow.