
Video MCP server: YouTube transcripts, summaries, chapters and timestamped answers for Claude
Video MCP server: YouTube transcripts, summaries, chapters and timestamped answers for Claude, Cursor and Codex. Hosted, no API key needed to start.
·
Give Claude, Cursor, Codex or any MCP client the context of a video from its link: the transcript, a summary, chapters, and the exact moments that answer a question, each with a timestamp link. Works on YouTube, TikTok, Instagram, X and Vimeo links. No account and no API key needed to start.
claude mcp add --transport http scribiz https://scribiz.com/mcpWatch the 48 second demo. Every response in it is a real call to the server.
This repo holds the setup files for the hosted Scribiz MCP server at https://scribiz.com/mcp. The server runs on scribiz.com. The Scribiz source code is not in this repo.
You paste a video link into a chat. The agent asks Scribiz for a short overview first. It searches for the part it needs, reads only that part, and answers with a link that opens the video at that moment. It does not read a whole transcript to answer one question.
The hosted server needs no account and no key. Add the URL and it works.
claude mcp add --transport http scribiz https://scribiz.com/mcpRun /mcp inside Claude Code to see the server and its tools.
Go to Customize, then Connectors. Select + Add, then Add custom connector. Enter:
Scribizhttps://scribiz.com/mcpThe claude_desktop_config.json file is a separate setup for local servers. You do not need it for the hosted server.
Put this in ~/.cursor/mcp.json, or in .cursor/mcp.json inside a project:
{
"mcpServers": {
"scribiz": { "url": "https://scribiz.com/mcp" }
}
}Put this in .vscode/mcp.json. VS Code uses servers, not mcpServers.
{
"servers": {
"scribiz": { "type": "http", "url": "https://scribiz.com/mcp" }
}
}Add this to ~/.codex/config.toml:
[mcp_servers.scribiz]
url = "https://scribiz.com/mcp"This repo is also a Claude Code plugin marketplace. The plugin adds the server and a skill that teaches Claude how to use it.
claude plugin marketplace add Illyism/scribiz-mcp
claude plugin install scribiz@scribiz-mcpUse the plugin or the claude mcp add line above, not both. Other agents can install the skill alone:
npx skills add Illyism/scribiz-mcpAsk your agent:
Summarize https://www.youtube.com/watch?v=jNQXAC9IVRw, then find where they talk about the trunks and quote it.This is a 19 second public video, "Me at the zoo". You should see a call to get_video_context, then search_video, and an answer with a timestamp link. The result says how the transcript was made. When a model read the link, its times are approximate (about 2 seconds either way). A video Scribiz has not processed before uses a little of the day's allowance on its first read, about 0.3 minutes for this one. After that it costs nothing.
All five tools are read-only. They reach out to the internet, and they never change your files.
| Tool | What it does |
|---|---|
get_video_context | Get video context. Title, length, summary, chapters and key moments with links. Start here. |
search_video | Search a video. The moments where a word or phrase is said or shown, ranked, with links. |
ask_video | Ask a question about a video. An answer with 3 to 5 cited moments. |
get_transcript | Get a video transcript. The words, or one part of them, one page at a time. In TXT, Markdown, SRT, VTT or JSON. |
get_job | Check a video job. Picks up a video that is still being processed. |
Every result says which layers ran (captions, listening, watching, a model summary) and how many minutes it used. Details are in the tools reference.
This runs inside a free allowance: about 10 minutes of reading a day for each connection (the same 10 minutes as the free web tool), videos up to 15 minutes, 5 questions a day and 30 caption lookups a day. Everyone who connects without a key also shares one more daily limit. When your 10 minutes or the shared limit are used up, captions and videos already processed still work. When the 30 caption lookups are used up, videos already processed still work. When the 5 questions are used up, searching and reading still work. The error says what is used up and when it resets (00:00 UTC).
Without a key the server never looks at the picture, so there are no on-screen notes, and a question about what was shown is answered from the words only. Watch reads the picture and the text shown on screen. It needs a Scribiz API key and uses minutes from your plan. A free account has 30 minutes a month, and you make a key in the dashboard. With a key, pass watch: true to get_video_context or ask_video; on-screen notes Scribiz already stored come back at no cost. See connection modes and limits and cost.
The hosted server takes public links only. It cannot read files on your disk.
Send the key as a bearer token. In Claude Code:
claude mcp add --transport http scribiz https://scribiz.com/mcp --header "Authorization: Bearer YOUR_SCRIBIZ_KEY"A local server comes with the command-line tool. It runs on your machine and can read your files. It needs Node 24, ffmpeg and yt-dlp, and has been tested on macOS only.
npm install -g scribiz
scribiz login
claude mcp add scribiz -- scribiz mcpYour agent sends the video link and your questions to scribiz.com. Scribiz fetches the captions from the site you linked when it can. For a video it has not processed, it passes the link to Google's Gemini API, whose model reads the public video. Nothing is uploaded from your machine. Where Scribiz downloads the audio, it sends the audio to Google for transcription and analysis. For Watch it also sends a small low-resolution copy of the picture, with no sound. The privacy policy has the retention periods and Google's terms. This repo sends nothing. It holds setup files only.
A video can say anything, including "ignore your instructions". The server wraps all text from a video between === BEGIN UNTRUSTED VIDEO TEXT [id] === and === END UNTRUSTED VIDEO TEXT [id] === lines, and sets untrusted_content: true on the result. Do not auto-approve tools that can send messages, run commands or write files in a session that reads video. More in MCP security.
Scribiz also has a free web tool. Paste a video link, get the transcript. No account needed: scribiz.com. There is a Mac app for Apple silicon Macs at scribiz.com/download, and a command-line tool on npm (npm install -g scribiz).
Built by Ilias Ism.
Pick your client and paste the snippet. Each one is the same server, written the way that client expects it.
claude mcp add scribiz --transport http https://scribiz.com/mcp{
"mcpServers": {
"scribiz": {
"url": "https://scribiz.com/mcp"
}
}
}code --add-mcp '{"name":"scribiz","url":"https://scribiz.com/mcp"}'Runs as a hosted service. Your client connects over the network and nothing is installed locally.
Paste this prompt into your agent. It reads this page and does the setup for you.
Read https://aiagentslisting.com/mcp/scribiz-mcp to learn what the "Scribiz Mcp" MCP server does and how to install it. Add it to my coding agent's MCP configuration as documented on that page, then confirm the server connects and list the tools it exposes.Agents can also browse this directory over MCP at https://aiagentslisting.com/api/mcp
claude mcp add --transport http scribiz https://scribiz.com/mcp
claude mcp add --transport http scribiz https://scribiz.com/mcp --header "Authorization: Bearer YOUR_SCRIBIZ_KEY"
claude mcp add scribiz -- scribiz mcpnpx skills add Illyism/scribiz-mcpnpm install -g scribiz| Name | What it does |
|---|---|
get_transcript | Read the transcript of a video, one page at a time. Use it only when you need the words themselves (quote, translate, copy, review). To answer a question or find where something is said, use ask_video |
get_video_context | Start here for any video. Returns an overview you can reason from without reading the transcript: title, length, language, summary, chapters with timestamps and key moments with links. detail "brief" |
ask_video | Ask a question about one video and get an answer with 3 to 5 cited moments (timestamps and links that open the video there). Use it for questions about the whole video or a topic across it, and for wh |
search_video | Find where something is said or shown in a video: ranked moments with timestamps and links, from the transcript, the on-screen notes, the chapters and the key moments. It is a local keyword search ove |
get_job | Check a video run that returned status "processing". Pass its job_id exactly as given (it can be long: copy all of it). It waits up to the server's wait time for the run to finish, then returns exactl |
Read live from the server on November 15, 2026.
Maintain this server? Claiming proves you own the listing and earns the last 4 points.
Claim this listingPaste this line near the top of your repository's README. The badge always renders the current score.
[](https://aiagentslisting.com/mcp/scribiz-mcp)Nothing comparable is listed yet.
No tagged releases on record.
Connect to the AI Agents Listing MCP server at https://aiagentslisting.com/api/mcp using the HTTP transport and complete the sign-in when it asks. Then call the submit_review tool for the listing "scribiz-mcp" with a rating from 1 to 5 and a short note on what worked and what did not.No reviews yet
Be the first to share your experience with this tool.
One email a week. New agents, MCP servers and skills, and what is actually getting traction.