MCP YouTube Transcript
An MCP server retrieving transcripts of YouTube videos
作者提供的说明(英文原文)
SEO for AI agents: keywords, SERPs, backlinks, rank tracking, audits, Search Console, Analytics
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。
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 …”。
适用于网页版、桌面版和手机版 Claude:
https://app.harborrank.com/mcp在终端中运行这条命令
claude mcp add --transport http harborrank https://app.harborrank.com/mcp粘贴到项目中的 .cursor/mcp.json,或 ~/.cursor/mcp.json
{
"mcpServers": {
"harborrank": {
"url": "https://app.harborrank.com/mcp"
}
}
}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": {
"harborrank": {
"type": "http",
"url": "https://app.harborrank.com/mcp"
}
}
}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。
ChatGPT 只能连接互联网上的服务器:
https://app.harborrank.com/mcp菜单名称可能因套餐和版本不同而略有差异。
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。
我们还没有读取这个服务器的工具列表;我们会轮流检查各个端点。
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6日
注册名称已通过域名 harborrank.com 验证(注册中心会核实域名所有权)
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
尚未检查;我们会轮流检查各个端点
尚未检查;我们会轮流检查各个端点
尚未检查;我们会在检查端点时读取工具列表
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。
注册中心没有列出必需的密钥,不过有些服务器会在首次使用时要求你登录。
它在作者的云端运行。通过互联网连接,无需安装,但你发送的内容会经过作者的服务器。
GitHub 仓库最后更新于–,最新版本(1.2.0)于20天前进入注册中心。
支持 MCP 的应用,例如 Claude Desktop、Claude Code、Cursor 和 VS Code。它通过 URL 连接,所以接受远程 MCP 服务器的应用也可以使用,例如开启开发者模式的 ChatGPT。
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。
Eighteen free, MIT-licensed Claude SEO skills, plus the HarborRank SEO MCP server that gives them live keyword, SERP, backlink and Google Search Console data. They work in Claude, Claude Code, Cursor, Gemini CLI, Codex, and other agents that support SKILL.md files. Every skill works on its own from web evidence and your files, and gets live numbers when HarborRank is connected.
Contributions welcome. Found a way to improve a skill or have a new SEO workflow to add? See CONTRIBUTING.md and open a PR.
Each skill is one slash command. With HarborRank connected, the skills pull live data through the HarborRank MCP server: keyword metrics, SERPs, ranked keywords, backlinks, rank tracking, Google Search Console, and Google Analytics.
If you ask before connecting, Claude says HarborRank isn't connected, gives these steps, and picks the task back up once you say "connected".
One install adds the HarborRank MCP server and all eighteen skills. Inside Claude Code:
/plugin marketplace add winsley-jeune/harborrank-skills
/plugin install harborrank@harborrank
Then run /mcp, pick harborrank, and sign in (setup guide). The first MCP call opens the HarborRank login in your browser. Installed this way the skills run as /harborrank: , for example /harborrank:keyword-research.
Install HarborRank SEO from the Cursor Marketplace, or add the MCP server by hand in ~/.cursor/mcp.json:
{ "mcpServers": { "harborrank": { "url": "https://app.harborrank.com/mcp" } } }
Cursor opens the HarborRank login the first time it connects. The plugin also adds the eighteen skills.
Add HarborRank from ChatGPT's plugin directory and sign in when ChatGPT asks. It connects the same MCP server; the skills in this repository are for agents that read SKILL.md files.
gemini extensions install https://github.com/winsley-jeune/harborrank-skills
Then run /mcp auth harborrank in Gemini CLI to sign in. The extension adds the MCP server and the eighteen skills.
SKILL.md files and can connect to a remote HTTP MCP server)./pricing indexed? If not, why not, and which of my pages are sitting at positions 8 to 20?".mcp.json (Claude Code), mcp.json (Cursor) and gemini-extension.json (Gemini CLI) all point at one remote MCP server, https://app.harborrank.com/mcp, over HTTPS with OAuth 2.0. No credentials are stored in this repository.For Codex or any agent that reads SKILL.md files, or if you already added the MCP server by hand:
npx skills add winsley-jeune/harborrank-skills
Install everything for Claude Code only:
npx skills add winsley-jeune/harborrank-skills --skill '*' --agent claude-code
Or copy the folders under skills/ into ~/.claude/skills/ (Claude Code) or ~/.codex/skills/ (Codex).
If you install the files only, connect HarborRank MCP first. The skill files are instructions, not data: without the MCP connection your agent only has what it can see on its own.
| Command | What it does |
|---|---|
/seo-project-setup | Create a local workspace with goals, competitors, exports, and preferences that later sessions reuse. |
/seo-context | Write the one-page context file (site, market, audience, competitors, goals) that every other skill reads first. |
/seo-coach | Pick the next workflow when you are new to SEO or unsure what to run first. |
/keyword-research | Turn seed topics into a prioritized keyword shortlist with volume, difficulty, intent, and SERP caveats. |
/keyword-clustering | Group a keyword list by intent, map clusters to pages, and flag cannibalization. |
/competitive-landscape | Map who wins a market, why, and where the openings are. |
/competitor-analysis | Study one competitor's keywords, pages, and backlinks and turn it into takeaways. |
/seo-audit | Crawl the site for technical issues, check indexing, and find why traffic dropped, with fixes ranked by impact (the crawl and Search Console checks use no credits). |
/content-brief | Turn one target keyword into a brief built from the live SERP: questions to answer, what to add that page one lacks, outline, links, schema and title. |
/local-seo | Check where a local business ranks in Google Maps across its service area, compare it with the businesses above it, and get a local action plan. |
/client-seo-report | Write a monthly SEO report for a client or boss from Search Console, Analytics and rank tracking, and publish it as a shareable link. |
/keyword-cannibalization | Find queries where two of your pages compete in Search Console, fix only the splits that cost traffic, and check Google's chosen canonical. |
/google-business-profile | Audit a Google Business Profile against the Maps leaders: categories, services, description, reviews, Q&A, NAP and landing page, with ready-to-paste text. |
/schema-markup | Find the structured data each page type has, write JSON-LD that matches the visible content, and validate it. |
/ai-search-visibility | See how often AI assistants send visits and cite the site, which sources they cite instead, and what to change to get cited. |
/seo-proposal | Turn a prospect's teardown into an SEO proposal: current state with numbers, opportunity ranges, a 90-day plan, and a cover email. |
/link-prospecting | Find qualified link prospects and the angle that makes each one relevant. |
/lead-teardown | Prospect local businesses in a trade and city, research each one, and get a one-page teardown to send as the pitch (research needs a paid plan). |
Docs for every skill: harborrank.com/docs/skills. What the skills are for and how they compare: Claude SEO skills built on live search data.
MIT. These skills started from the open-source OpenSEO project's skills and keep its license.