MCP YouTube Transcript
An MCP server retrieving transcripts of YouTube videos
作者提供的说明(英文原文)
Make faceless YouTube videos and Shorts from a topic or script, then publish to YouTube.
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://narva.app/mcp在终端中运行这条命令
claude mcp add --transport http narva https://narva.app/mcp粘贴到项目中的 .cursor/mcp.json,或 ~/.cursor/mcp.json
{
"mcpServers": {
"narva": {
"url": "https://narva.app/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": {
"narva": {
"type": "http",
"url": "https://narva.app/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://narva.app/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。
服务器要求先登录,才会列出它的工具。
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 narva.app 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
在 589 毫秒内响应(HTTP 401)
声明支持 OAuth 登录,你的 AI 应用可以带你去登录
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 3 项适用的基础检查中通过了 3 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(1.0.1)于4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。
Narva makes faceless YouTube videos: give it a topic, a finished script or a recorded voiceover and it delivers a ready-to-upload video or Short, with the script, natural AI narration in 32 languages, AI visuals in 12 styles with camera moves, real archive footage, animated maps and charts, word-synced captions, music and thumbnails. Then it publishes to your YouTube channel, now or scheduled.
This repository holds the public manifests for Narva's hosted MCP server, so Claude, ChatGPT, Cursor or any MCP client can make and publish videos for you mid-conversation.
https://narva.app/mcp (Streamable HTTP)Try: *"Make a 10-minute documentary about the Great Emu War in a dark documentary style and publish it to my channel tomorrow at 6pm."*
Claude.ai (web or desktop): Settings → Connectors → Add custom connector → paste the server URL and approve the consent screen.
Claude Code:
claude mcp add --transport http narva https://narva.app/mcp
Or install this repository as a plugin, which adds the server together with a skill that teaches Claude the Narva workflow:
claude plugin marketplace add Negmus/narva-mcp
claude plugin install narva@narva-mcp
Settings → Security and login → turn on Developer mode, then Plugins → + → paste the server URL. The OAuth prompt signs you in to Narva.
Add to Cursor, or in ~/.cursor/mcp.json:
{ "mcpServers": { "narva": { "url": "https://narva.app/mcp" } } }
code --add-mcp '{"name":"narva","type":"http","url":"https://narva.app/mcp"}'
Install this repository as a plugin, which adds the server together with the same skill:
codex plugin marketplace add Negmus/narva-mcp
codex plugin add narva@narva codex mcp login narva
Or only the server, in ~/.codex/config.toml:
[mcp_servers.narva]
url = "https://narva.app/mcp"
then codex mcp login narva to approve the consent screen.
gemini extensions install https://github.com/Negmus/narva-mcp
Sign in with /mcp auth narva.
In mcp_config.json (Agent panel → MCP Servers → Manage → View raw config):
{ "mcpServers": { "narva": { "serverUrl": "https://narva.app/mcp" } } }
Or install this repository as a plugin: agy plugin install https://github.com/Negmus/narva-mcp
In ~/.codeium/windsurf/mcp_config.json:
{ "mcpServers": { "narva": { "serverUrl": "https://narva.app/mcp" } } }
Add to Kiro, or in .kiro/settings/mcp.json:
{ "mcpServers": { "narva": { "url": "https://narva.app/mcp" } } }
MCP Servers → Remote Servers → name narva, URL https://narva.app/mcp, transport Streamable HTTP.
goose configure → Add Extension → Remote Extension (Streamable HTTP) → https://narva.app/mcp.
Add to LM Studio
The OAuth callback cannot reach a client that runs on a server, so create a personal access token under Settings → API & MCP on narva.app and send it as a header:
mcp_servers:
narva: url: https://narva.app/mcp headers: Authorization: Bearer YOUR_TOKEN
The same token works with the REST API (https://narva.app/api/v1).
Everything is scoped to the signed-in user's own account and videos.
| Tool | Kind | What it does |
|---|---|---|
get_account | read | Credits, plan, price per minute and connected YouTube channels |
list_options | read | Formats, niches, tones, languages, visual and caption styles, music moods, voices |
list_videos / get_video | read | Your videos with status, progress, cost, script, scenes, download and editor links |
list_youtube_channels | read | Channels connected for publishing |
create_video | write | Start a video from a topic or a script; on autopilot it renders by itself |
update_script | write | Replace the script of a video that has not been rendered yet |
plan_video | write | Cut the script into scenes and direct them; shows the exact credit cost |
render_video | write | Render a planned video (charges the shown credits; failures are refunded) |
edit_scene | write | Change one scene's line, image or motion graphic |
edit_video | write | Describe a change in words; Narva proposes the edits, then applies them |
create_short_from_video | write | Turn a finished long video into a 9:16 Short |
publish_to_youtube | write | Upload to your channel now or scheduled, with title, description, tags and thumbnail |
delete_video | write | Delete a video and its files (not copies already on YouTube) |
Every tool carries readOnlyHint, destructiveHint, openWorldHint and idempotentHint annotations. Rendering spends credits and publishing posts public content, so the server tells the assistant to confirm both with you first.
server.json: the entry in the official MCP Registry as app.narva/mcp.claude-plugin/ + .mcp.json + skills/narva/SKILL.md: a Claude Code plugin bundling the server with a skillplugin.json + mcp.json: the same as an Agent Plugin for Codex (OpenAI settings under extensions.com.openai), Cursor and compatible clients; .agents/plugins/marketplace.json makes the repository a Codex marketplacegemini-extension.json: a Gemini CLI extension.antigravity-plugin/plugin.json + mcp_config.json: an Antigravity plugin; plugins/narva/ holds the same without the skillSETUP.md: the connection walkthrough a client can follow after installThe server itself is part of the Narva web application and is not open source. The manifests and skill here are MIT licensed.
The server only reads and writes the signed-in user's own Narva data, and publishes to YouTube only the channels the user connected. Privacy policy: https://narva.app/privacy. Terms: https://narva.app/terms. Support: support@narva.app.