Skip to content
Get packs

Guide · Agent skill

How to use the Video Transcript Downloader agent skill

Updated · written with AI from the project's own documentation and checked against our catalogue data

The Video Transcript Downloader is an agent skill from steipete/agent-scripts (MIT licence) that lets an AI agent pull a text transcript from a video link and download video, audio or subtitle files using yt-dlp. This guide explains what agent skills are, how to install this one, how to ask for what you need, and what it requires. By the end you will be able to set it up and request a transcript or download in plain language.

Video Transcript Downloadersteipete/agent-scripts · MIT

Quick start

  1. 1Download and unzip the skill. Download the ZIP of the skill from our page and unzip it. Keep the folder name video-transcript-downloader.
  2. 2Place it in a skills folder. Move the folder to ~/.claude/skills/ for personal use, or to .claude/skills/ inside a project. Alternatively, upload it under Settings → Capabilities → Skills in the Claude apps.
  3. 3Install Node.js dependencies. Run npm ci in the skill folder to install the packages from package-lock.json.
  4. 4Install yt-dlp and ffmpeg. Install both tools, for example with brew install yt-dlp ffmpeg on macOS, then confirm them with yt-dlp --version and ffmpeg -version.
  5. 5Ask for what you need. Give your agent a video link and a plain request, such as a transcript, subtitles or an audio download.
  6. 6Check the output. Read the returned transcript or confirm the file in your output folder, and check that the video is one you are allowed to use.

What does the Video Transcript Downloader skill make the agent do?

Once installed, the agent can run a bundled Node.js command-line tool, scripts/vtd.js, to print a transcript of a video as a clean paragraph. Timestamps are optional. The tool can also download the video, the audio track or subtitle files.

For YouTube, transcripts are fetched with the youtube-transcript-plus package when possible. For other sites, the tool pulls subtitles through yt-dlp and cleans them into a paragraph. The tool can also list available video formats and pass extra options through to yt-dlp.

  • Transcript as a single readable paragraph, with or without timestamps
  • Download video, audio or subtitles to a folder you choose
  • List available formats and download a specific format ID
  • Pass extra yt-dlp options through the agent

Install the skill: download the ZIP and unzip it

Download the ZIP of this skill from our page and unzip it. Then place the folder in one of two locations: your personal skills folder at ~/.claude/skills/, or the .claude/skills/ folder inside a specific project.

Alternatively, in the Claude apps, go to Settings → Capabilities → Skills and upload the skill there.

  • Personal: ~/.claude/skills/video-transcript-downloader
  • Project: .claude/skills/video-transcript-downloader inside your project folder
  • Claude apps: Settings → Capabilities → Skills, then upload

Install the tool's dependencies before first use

The skill runs scripts, so its Node.js dependencies must be installed. The source sets up the skill from its folder inside the agent-scripts repository, using npm ci, which installs the exact packages listed in package-lock.json.

You also need yt-dlp and ffmpeg. The source gives this command on macOS with Homebrew, and a command to check both tools are working.

  • Node.js and npm must be installed
  • Run npm ci in the skill folder so the packages from package-lock.json are installed
  • Install yt-dlp and ffmpeg, then check them with yt-dlp --version and ffmpeg -version
cd ~/Projects/agent-scripts/skills/video-transcript-downloader && npm ci brew install yt-dlp ffmpeg yt-dlp --version ffmpeg -version | head -n 1

Ask for a transcript in natural language

You do not need to type the command yourself. Describe what you want, give the video link, and the agent chooses the right subcommand. The source defines the default transcript as a single paragraph, and says timestamps should be used only when asked.

Behind the scenes, the agent can run commands such as the ones below. You can also run them yourself to check the tool works.

  • "Get the transcript of this lecture: [link]"
  • "Give me the transcript in English with timestamps: [link]"
  • "Keep the [Music] markers in the transcript of this interview: [link]"
./scripts/vtd.js transcript --url 'https://…' ./scripts/vtd.js transcript --url 'https://…' --lang en ./scripts/vtd.js transcript --url 'https://…' --timestamps ./scripts/vtd.js transcript --url 'https://…' --keep-brackets

Download video, audio or subtitles

Ask the agent to save a video, an audio track or subtitle files to a folder. The source uses ~/Downloads as the example output folder. Only do this for media you have the right to use.

To choose a specific format, ask the agent to list the formats first. Extra yt-dlp options can be added after a double dash, and the source includes -v as an example for verbose output.

  • "Download this video to my Downloads folder: [link]"
  • "Save only the audio of this talk: [link]"
  • "Download English subtitles for this video: [link]"
  • "List the available formats for this link"
./scripts/vtd.js download --url 'https://…' --output-dir ~/Downloads ./scripts/vtd.js audio --url 'https://…' --output-dir ~/Downloads ./scripts/vtd.js subs --url 'https://…' --output-dir ~/Downloads --lang en ./scripts/vtd.js formats --url 'https://…'

Safety, network and limits to know

This skill runs scripts and uses the network. It fetches media from video sites, so the terms of those sites and copyright law may restrict what you can download. Use it only for content you have rights to.

The tool has no credential handling. Private videos behind a login may give incomplete results. Setup will fail if yt-dlp, ffmpeg, Node.js or npm are missing.

  • Runs scripts and uses the network
  • Use only for videos you are allowed to use
  • No login or credential handling for private or sensitive videos
  • Review the bundled files (scripts/vtd.js, scripts/html-entities.js, scripts/vtd.test.mjs, package.json, package-lock.json) if you want to inspect what runs

What people use it for

  • Turning a recorded lecture or talk into text for study notes without watching it again
  • Getting a quick transcript or subtitle file from a video you created or are licensed to use
  • Saving an audio track of a podcast or talk for offline listening where permitted
  • Checking which video formats a link offers before choosing one to download

Frequently asked questions

How do I install the video transcript downloader agent skill?

Download the skill ZIP from this page and unzip it. Put the folder in ~/.claude/skills/ for personal use, or in .claude/skills/ inside a project. You can also upload it in the Claude apps under Settings → Capabilities → Skills.

What do I need installed to use it?

You need Node.js and npm to run npm ci and the script, plus yt-dlp and ffmpeg. The npm packages come from package-lock.json, including youtube-transcript-plus. Setup fails until these are present.

Is the Video Transcript Downloader skill free?

The skill is published under the MIT licence, as stated on this page. Check the licence file in the source repository for the full terms before reuse.

Does it work with YouTube and other sites?

YouTube transcripts are fetched directly with youtube-transcript-plus when possible. For other sites, the tool pulls subtitles through yt-dlp and cleans them into a paragraph.

Can it download videos I do not own?

The tool fetches media from video sites, and the terms of those sites and copyright law may forbid downloading content you have no rights to. Only use it for videos you are allowed to save.

Why does the transcript fail or come back incomplete?

Check that yt-dlp and ffmpeg are installed and working, and that npm ci has run. Private or login-only videos may give incomplete results because the tool has no credential handling.

New agent skills, repos and AI tools — every week

The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.

Our verdict

This skill suits researchers, students and creators who want transcripts, subtitles or offline copies of videos they are allowed to use, and who are comfortable installing Node.js, yt-dlp and ffmpeg. It is not a good fit for copyrighted material you have no rights to, for private videos behind logins, or for anyone who cannot install command-line tools.

Download the skill