Vidskim: YouTube transcripts

com.vidskimPublished by vidskim.comOfficial publisherv1.0.2Updated Oct 31, 2026

Let AI assistants read YouTube: transcripts with clickable timestamps, chapters, search, channels.

VerifiedStreamable HTTPWeb executableWeb Search & ScrapingMedia & Design

Overview

AI-generated overview

Hosted MCP server that lets an assistant read public YouTube videos: full transcripts with clickable timestamps, chapters, in-video search, and channel or…

What you can do
  • Retrieve a full transcript in Markdown with chapters and a timestamp link on every paragraph, in parts or by time range, with optional machine translation
  • Search inside a single video for passages about a topic
  • Look up video details: title, channel, date, views, description and chapters
  • Search YouTube by topic and list a channel's latest uploads or the videos in a playlist
When to use it
Useful when you want an assistant to answer from what a YouTube video actually says, with timestamps you can check, or to find videos and channel uploads before you have a link. Suits research, summarising talks or tutorials, and quoting specific moments.
Requirements
Remote Streamable HTTP endpoint; nothing to install and no API key. Sign-in with Google through OAuth 2.1 with dynamic client registration, added as a custom connector or remote MCP server in the client. Only public videos with captions work; the free plan covers 30 videos a month.
Before you install
Signing in with Google grants the server an OAuth authorization, so review what you allow. All tools are read-only and the server does not post to or touch your YouTube account. Transcripts are speech only, so on-screen text and slides are missed, and videos without captions, private or members-only videos return a message instead of content. Usage is capped at 30 videos a month on the free plan.

Installation

In SourceWeft

  1. Open Vidskim: YouTube transcripts in the dashboard and add it to a workspace.
  2. Connect the required credentials. They are stored privately and never shown on public pages.
  3. Enable the server for the chats that should use its tools.

Web executable via Streamable HTTP. Remote servers run from the web runtime once configured in a workspace.

Use with an agent

Paste this prompt into Claude Code, Codex, Cursor or another agent — or into SourceWeft chat. The agent reads this server's install guide, shows you its publisher, endpoint and sign-in, and adds it to your client once you agree. Keys never go into a command.

Read https://sourceweft.com/mcp/com.vidskim%2Fyoutube/install.md and connect the MCP server it describes to my client. Before changing any config, show me its publisher, endpoint and sign-in method and wait for my OK. Never put an API key in a command or in a file I would commit; ask me where it should come from.

Read the install guide the agent follows

Other MCP clients

Add this to your client's mcpServers config. The client completes OAuth sign-in on first connect.

{
  "mcpServers": {
    "vidskim-youtube-transcripts": {
      "type": "http",
      "url": "https://vidskim.com/api/mcp"
    }
  }
}

README

[Vidskim logo]

Vidskim: YouTube MCP server

A hosted Model Context Protocol server that lets Claude, ChatGPT, Claude Code, Cursor, VS Code, and other MCP clients read YouTube videos. It returns full transcripts with chapters and a clickable timestamp on every paragraph, so the assistant answers from what was actually said and you can check each claim in one click.

https://vidskim.com/api/mcp

Remote server, Streamable HTTP, OAuth 2.1 with dynamic client registration. Nothing to install, no API key: paste the URL and sign in with Google.

Website and docs: vidskim.com/mcp · Registry: com.vidskim/youtube · Listed in the Claude connector directory

Tools

ToolWhat it does
get_transcriptFull transcript in Markdown with chapters and a timestamp link on every paragraph. Long videos come in parts; any time range can be read on its own; YouTube's machine translation is available for other languages.
search_transcriptFinds the passages about a topic inside one video.
get_video_infoTitle, channel, date, views, description, and chapters.
search_youtubeFinds videos by topic when you don't have a link yet.
list_channel_videosA channel's latest uploads, or the videos in a playlist.

All tools are read-only. Vidskim never posts, comments, or touches your YouTube account.

Connect

Claude (claude.ai, Desktop, mobile)

Vidskim is listed in the Claude connector directory. Open claude.ai/directory/vidskim, click Connect to Claude, sign in with Google, and press Allow.

Or add it as a custom connector:

  1. Open Settings → Connectors and choose Add custom connector.
  2. Name it Vidskim and paste https://vidskim.com/api/mcp.
  3. Click Connect, sign in with Google, and press Allow.

Tip: choose Always allow for the Vidskim tools the first time Claude asks, so long videos (which arrive in several parts) don't ask again for every part.

ChatGPT

  1. Open Settings → Apps & Connectors → Advanced and turn on Developer mode.
  2. Back in Apps & Connectors, click Create, name it Vidskim, paste the URL, and pick OAuth.
  3. Sign in with Google, press Allow, then turn Vidskim on in a chat from the + menu.

Claude Code

bash
claude mcp add --transport http vidskim https://vidskim.com/api/mcp

Then type /mcp inside Claude Code, pick vidskim, and sign in.

Cursor

Add to ~/.cursor/mcp.json:

json
{  "mcpServers": {    "vidskim": { "url": "https://vidskim.com/api/mcp" }  }}

Cursor opens the sign-in the first time a tool runs.

VS Code

bash
code --add-mcp '{"name":"vidskim","type":"http","url":"https://vidskim.com/api/mcp"}'

Start the server from the MCP view and sign in when Copilot asks.

Any other MCP client

Add a remote MCP server with the Streamable HTTP transport and the URL above. Authentication is standard OAuth 2.1 with dynamic client registration.

Example

Explain how this video says a transformer works, in a few short points with timestamps: https://youtu.be/wjZofJX0v4M

  • A chatbot works by predicting a probability distribution for the next chunk of text, sampling from it, and repeating 2:01.
  • The input is split into tokens, and each token becomes a vector meant to encode its meaning 3:04.
  • In the attention block those vectors pass information to each other 4:05.

Limits

  • Works with public YouTube videos that have captions, uploaded or auto-generated, in any language. Videos with no captions, private, members-only, or removed videos return a clear message instead of a guess.
  • A transcript is speech only: slides and on-screen text are missing unless the speaker reads them out.
  • Free plan: 30 videos a month. Each video counts once, so follow-up questions and searches inside it don't use more.

Privacy and support

This repository holds the listing files for MCP directories (this README and server.json). The server itself is hosted at vidskim.com.

Vidskim is independent and is not affiliated with, endorsed by, or sponsored by YouTube, Google, Anthropic, or OpenAI.

Source: README.md at commit 789625d

Tools

0
This server asks you to sign in, so SourceWeft could not read its tools without an account and has not indexed them. What a connected account sees can depend on that account and its permissions.

Version history

2
  1. v1.0.2LatestOct 31, 2026
  2. v1.0.1Oct 29, 2026

Score

85% of the applicable checks met

How complete this listing is, computed from the registry entry, the endpoint and public sources. Not a rating of the server and not a statement that it is safe or working today: reachability is a dated check, and no installs or reviews go into the score.

  • Reached in the last 30 daysRequired25 pts
  • Auth declaredRequired10 pts
  • Tools listed15 pts

    Answer tools/list without a credential, or ship a package that starts in a sandbox without one.

  • DescriptionRequired15 pts
  • Logo10 pts
  • Documentation15 pts
  • License10 pts