08:31Brad | AI & Automation · Tutorial
This Skill Can INSTANTLY Watch any Video For Free — Here's How
A Claude Code plugin update that makes an AI watch any video 40 times faster by grabbing keyframes instead of decoding every frame.
July 2ndThe argument in one line.
Claude can now watch and analyze any video by converting it into frames and timestamped transcripts instead of relying on expensive video models, costing roughly $1 per video and taking minutes instead of hours of manual scrubbing.
Read if. Skip if.
READ IF YOU ARE…
- A content creator or researcher who regularly watches long-form videos (30+ minutes) and currently spends hours manually taking notes or scrubbing for specific moments.
- An AI enthusiast with basic CLI familiarity who wants to integrate video analysis into existing Claude workflows without building from scratch.
- A product person or founder who needs to extract key insights, decisions, or visual context from recorded talks, lectures, or demos quickly and cheaply.
SKIP IF…
- You primarily work with video content that has no transcript or subtitles available, since the skill relies on YouTube captions or existing subtitle files.
- You need real-time video analysis or live-stream processing — this tool works with finished, uploadable video files only.
The full version, fast.
A free, open-source Claude Code skill called /watch gives Claude the ability to ingest any video � YouTube, Loom, Instagram reels, local files, anything yt-dlp supports � by treating the video as two things Claude already understands: a sequence of frames and a timestamped transcript. The pipeline runs locally with yt-dlp and ffmpeg, pulls free YouTube captions when available and falls back to Whisper on Groq's free tier when not, and caps long videos at 100 frames so a 45-minute lecture and a one-hour video both cost about a dollar. The result is full visual plus verbal context, unlocking instant content research, hook breakdowns, screen-recording debugging, and an automatically fed Obsidian second brain.
Free for members
Chat with this breakdown — free.
Sign in and you get 23 free chat messages on us — ask for the hook, quote a framework, find the exact transcript moment, generate a markdown action plan. Bring your own key when you want unlimited.
Create a free account →Where the time goes.
00:00 – 00:52
01 · Cold open
Problem stated: other transcript tools only read words and miss half the video. Promise: how it works, the life-changing use case, and a 5-minute setup.
00:52 – 02:43
02 · Watch videos in minutes — live demo
Side-by-side screen recording: 45-minute Sam Altman YC lecture ingested in under 2 minutes. Claude returns structured speaker summary, queryable in terminal.
02:43 – 03:43
03 · Setup
GitHub link (free), install commands, automatic dependency install, API auth on a free-tier transcription service.
03:43 – 05:18
04 · Under the hood
Core insight: a video is just two things — frames and a transcript. yt-dlp + FFmpeg do the heavy lifting locally. No MCP, no third-party wrapper, no cloud service.
05:18 – 06:30
05 · The cost math
Frame scaling table: 1 min = 60 frames / $0.70; 1 hr = 100 frames / $1.62 (capped). YouTube captions are free; Groq Whisper free tier covers everything else.
06:30 – 07:03
06 · Analyze video hooks
Use case #1: content research — paste a winning video URL, ask Claude to break down the hook. Replaces 10 min/video of manual scrubbing.
07:03 – 07:42
07 · Debug screen recordings
Use case #2: developer QA — drop in a 30-second screen recording of a UI bug; Claude pinpoints the exact frame the state change happens.
07:42 – 08:36
08 · Content intelligence / second brain
Use case #3: Obsidian second brain — Claude auto-watches competitor videos and feeds structured notes in. Compounds over time.
Lines worth screenshotting.
- A video is just frames plus a transcript — splitting it into those two pieces lets Claude process it using capabilities it already has: reading images and text.
- yt-dlp and FFmpeg are the battle-tested tools every video platform already uses under the hood — combining them locally eliminates every third-party dependency.
- Frame-by-frame screenshots paired with a per-second timestamped transcript give Claude the full on-screen context that transcript-only tools miss entirely.
- Half of the interesting information in most videos happens on screen and is never said aloud in the transcript — reading only the transcript means missing half the content.
- A 45-minute YC lecture is fully ingested, summarized, and queryable in under two minutes using this pipeline.
- The skill caps frame extraction at 100 frames for videos over 30 minutes, which means a 30-minute video and a 60-minute video cost roughly the same token budget.
- YouTube auto-captions are free and the skill pulls them directly — Whisper only runs on content without captions, like local MP4 files or Looms.
- Groq's free tier provides two hours of Whisper transcription per hour — more than most users will realistically hit.
- Running five hours of video watched live by Claude burned less than 10% of one session's token budget.
- A Claude skill that gives the model video comprehension is a context expansion primitive — everything Claude could do with documents it can now do with video.
- The matrix moment is when Claude has already finished analyzing a 45-minute video before the video player reaches the two-minute mark.
- Sharing the video skill on GitHub for free and inviting optimization pull requests compounds the tool's value faster than building in isolation.
Give Claude Eyes — It Can Watch Any Video in Under Two Minutes
Claude Code skill
A free open-source Claude Code skill splits any video into frames and a transcript, letting Claude analyze a 45-minute lecture in under two minutes for roughly $1.62 per hour of footage.
01Cold open
- The problem with transcript tools is that they read words but miss half the video — combining frames with audio is the fix
- The promise: how it works, the life-changing use case, and five minutes to set it up
02Watch videos in minutes — live demo
- 45-minute Sam Altman YC lecture ingested in under two minutes — structured speaker summary, queryable in the terminal
- The side-by-side screen recording proves the speed claim before the explanation
04Under the hood
- A video is just frames plus a transcript — yt-dlp and FFmpeg extract both locally
- No MCP, no third-party wrapper, no cloud service — the processing runs on your machine with free-tier API calls
05The cost math
- One hour costs $1.62 at 100 frames — YouTube captions are free and Groq Whisper free tier covers transcription
- Frame count is capped so the cost does not scale linearly with video length
06Analyze video hooks
- Paste a winning video URL and ask Claude to break down the hook — replaces ten minutes of manual scrubbing per video
- Content research at scale becomes a query operation rather than a watch-and-note operation
07Debug screen recordings
- Drop a 30-second screen recording of a UI bug and Claude pinpoints the exact frame the state change happens
- Developer QA becomes a visual query rather than a frame-by-frame manual review
08Content intelligence / second brain
- Claude watches competitor videos automatically and feeds structured notes into Obsidian — the knowledge base builds without manual input
- The compounding effect is real — the longer it runs, the more context Claude has for any content decision
Terms worth knowing.
- Claude Code
- Anthropic's terminal-based coding agent that runs locally and can execute commands, read files, and call external tools through skills or MCP integrations.
- Claude skill
- A packaged set of instructions and scripts that extends Claude Code with a new capability, invoked with a slash command like /watch.
- Frame extraction
- Pulling still images from a video at fixed intervals so a language model can analyze the visual content one screenshot at a time.
- Timestamped transcript
- A text transcript where each line is tagged with the second it was spoken, letting a model align what was said with what was on screen.
- yt-dlp
- A free open-source command-line downloader that pulls video and audio from YouTube and roughly a thousand other sites, including captions when available.
- FFmpeg
- A long-standing free command-line tool for processing audio and video, used here to slice a clip into screenshots and extract a clean audio track.
- Whisper
- OpenAI's open-source speech-to-text model that converts audio into text, used as a fallback when a video has no pre-made captions.
- Groq
- A fast AI inference provider that hosts open models like Whisper at very low latency and offers a generous free tier for transcription.
- MCP
- Model Context Protocol — Anthropic's standard for connecting Claude to external tools and data sources through a running server, distinct from skills that run scripts directly on your machine.
- Context window
- The amount of text and images a language model can hold in working memory for a single conversation; exceeding it forces older content to drop out.
- Token budget
- The metered limit of input and output tokens a user can consume in a Claude session or billing period before hitting throttling or charges.
- Pattern interrupt
- A sudden shift in a video's pacing, visual, or message designed to break a viewer's autopilot scrolling and force them to pay attention.
- Hook
- The opening few seconds of a short-form video engineered to stop the scroll and convince a viewer to keep watching.
- Second brain
- A personal knowledge base — often built in a note-taking app like Obsidian — where ideas, clips, and references are stored and linked for later retrieval.
- Obsidian
- A local-first markdown note-taking app popular for building linked knowledge bases, with files stored as plain text on the user's machine.
- Loom
- A screen-recording service that produces shareable video links, commonly used for quick walkthroughs and async product demos.
Things they pointed at.
Lines you could clip.
01:30
“Half of the interesting stuff in a video isn't said out loud. It happens on screen.”
02:23
“You're not watching content anymore. You're actually downloading context automatically and putting it to work straight away.”
02:20
“That's the matrix moment.”
06:20
“I've used this skill every day for two weeks, and I'm still on the free tier. It's crazy.”
07:40
“Whatever you're using video for, you can probably stop watching it manually because of this skill.”
Word for word.
Read-along
Don't just watch it. Burn it in.
See every word as it's spoken — crank it to 2× and still catch all of it. The same dual-channel trick behind Amazon's Kindle + Audible.
metaphoranalogy
When you give Claude code the ability to instantly watch any video on the Internet for free, it becomes genuinely unstoppable. With this Claude skill, Claude can understand video as well as it reads PDFs.
Hours long YouTube videos, Instagram reels, looms, local files, anything. Before, Claude was just guessing.
Now it can watch the whole thing frame by frame instantly. It's like Neo plugging into the matrix. By the time you've hit play, Claude's already watched the whole thing and become an expert.
I've tried a bunch of transcript tools before developing this one, and they all let me down. They either cost way too much or they only ever read the transcript and missed half the video. This skill gives Claude the frames and the audio together, so it actually sees what's happening on screen.
Right now, I'll walk you through exactly how it all works, the use case that completely changed how I consume content, and how to set this up in your own Claude code in under five minutes. Here's what it actually looks like on a forty five minute video done in less than a few minutes. On the left, I have a YC lecture from Sam Altman about how to start a startup.
I'm gonna press play on that now and then grab the URL. All I have to do is go over to Claude and type slash watch and then paste the URL here. Then Claude gets to work grabbing the subtitles from YouTube for extracting the frames, and analyzing them all together.
So the reason this is better than just pulling the transcript is because Claude can actually grab the frames from this video. In this lecture, Sam goes through and shows a bunch of really great graphs. And this is important context for Claude because if you're only getting the transcript, you're only getting half of the information.
Now here's where most of the existing video tools fall short because they base everything around the transcript. When something happens on screen and it's not explicitly referenced in the transcript, Claude doesn't know about it, and you miss out on key context, which matters because half of the interesting stuff in a video isn't said out loud.
It happens on screen. So this skill actually watches. It pulls frame by frame screenshots and puts it together with a per second time stamped transcript to get Claude the full picture and full context.
And just like that, we're only two minutes into the lecture. Sam is still introducing what he's gonna talk about today, and Claude has already ingested the entire thing. I have a structured summary of all of the speakers.
I can see exactly what they talk about, and now I can actually query Claude on anything about this context and then start to put it to work instantly right here in the terminal. That's a forty five minute video done in less than two minutes watched, analyzed, and applied.
That's the matrix moment. You're not watching content anymore. You're actually and putting it to work straight away.
And you're probably thinking at the moment, there's some expensive API doing the heavy lifting here, but there isn't. But before we get into that, let's get into the setup. By the way, I'm giving this whole skill away for free on GitHub.
The link is in the description below. Just run these install commands and the setup takes care of the rest. Once the skill is installed, Claude runs the setup script and installs any dependencies that you don't have already.
It authenticates with the transcription API. Don't worry. This one is pretty much free and we'll get to it in a second.
But under the hood, the pipeline is actually surprisingly simple. Now here's the part that nobody really talks about. Claude can't actually watch video because Anthropic doesn't have a video model yet.
There are some other providers that can, like Google's Gemini model, but they're pretty expensive and they don't integrate nicely with Claude. So if you're watching a lot of content, that bill stacks up pretty fast. Luckily, there's a smarter way to do this because if you really break it down, a video is just two things.
It's a bunch of frames and a transcript. That's it. So instead of paying for another expensive model, I can just split the video into those two pieces and hand it to Claude in a format that it already knows how to read, pictures and text.
Now this is the part I love because the skill is doing this with two of the oldest, most battle tested line tools on the Internet, YouTube DLP and FFmpeg. These aren't MCPs. They're not some new wrapper.
There's no third party service involved in the middle here. They install one song right on your machine. Millions of developers have used them for over a decade now.
They're rock solid and completely free. And they're what every video tool you've ever touched is probably using under the hood. YTDLP is the downloader.
You can think of it as a right click save video, but it works on basically the whole Internet. FFmpeg is the video engine.
It takes the video and turns it into two things that Claude actually wants. First, screenshots, which are taken every few seconds all the way through the video, and then second, the audio file, which is pulled out as a clean little file ready to be transcribed into text using Whisper. Now Claude has the full picture when we put these two together.
It's flipping through the screenshots like a flipbook, reading the transcript like a script, and the time stamps line up exactly so it knows on screen when something is being said. So that's the whole pipeline. YouTube DLP and FFmpeg doing all the heavy lifting locally on your laptop for free.
The only thing we actually have to pay for here is the transcription and Claude usage. Caption's transcription is pretty much free. The skill just pulls them.
And if it doesn't, it transcribes the audio using Whisper hosted on Grok or OpenAI. I prefer Grok because it's extremely fast, and their free tier covers basically anything you throw at it. So most videos cost you literally nothing to transcribe.
I even used this exact skill to grow a universal context layer for content research, and I'll show you exactly how it works in a minute. And I can literally hear the keyboards clattering right now. Brad, this is gonna torch your token budget.
But this actually surprised me, so let's do the math. The skill scales frame count to video length, and it actually caps anything over thirty minutes to a 100 frames. So thirty minute video and a one hour video pretty much cost the same amount in dollar terms, and that's about $1 per run.
I ran every test in this video three times in parallel and burned less than 10% of my session, and that's over five hours of video watched live by Claude with transcription. And the transcription part's where it gets ridiculous. Every YouTube video comes with a free transcript.
The skill just pulls them. There's no Whisper, no API call. It's totally free.
And that goes for a bunch of other sites too. Whisper only kicks in for the stuff without captions, like a raw m p four, a Loom, or Instagram Reel. Grok's free tier actually gives you two hours of transcriptions per hour, which covers more than you'll realistically throw at it.
I've used this skill every day for two weeks, and I'm still on the free tier. It's crazy. Look.
I'm not saying this is perfect, and there's probably optimizations I haven't thought of, but for most people watching, this is essentially free. If you got ideas to make it cheaper or quicker, drop them in the comments below.
Once I realized this was basically free, I started running it on everything, which is how I ended up building the system I'll show you at the end, and it's one that's generally changed how I consume content. Here's the part that actually makes this skill a must have.
It works on any URL YTTLP supports, which is over a thousand sites. So this isn't just limited to the big social media companies or YouTube. And it even works if you have the files locally downloaded.
So that opens up a bunch of use cases that you probably wouldn't expect. So this is what I'm doing for content research now. I take a winning video from the Internet, and I ask Claude to break down the hook.
Claude tells me the visual setup, the exact words, where the pattern interrupt lands, and what's on screen at the moment of the hook. That used to take me ten minutes per video pausing and scrubbing, now it's just a paste. And for developers, there's another use case, debugging screen recordings.
If you're a developer and a UI bug shows up, you record a thirty second screen recording, drop it into Claude, and ask what happens right before the crash. Claude reads the frames around that moment, finds a state change, and tells you the exact frame the issue starts with. That alone has saved me hours.
The skill also has a zoom flag, start time and end time. So you can drop those in and click and focus frame by frame extraction on a specific window of a video. So you can ask about a ten second segment of a two hour video without burning your entire context window.
Whatever you're using video for, you can probably stop watching it manually because of this skill. So earlier, I told you that once you start using this thing, it seriously starts to change how you consume content. Now I wanna show you my personal favorite use case for it, which is feeding my second brain.
I keep a knowledge base in Obsidian with notes, snippets, ideas for content. And the bottleneck for me has always been throughput because there's just so much good content out there by creators at the moment. There's not enough time to watch it all and write it all down.
So I let Claude do both. I give it every single competitor that I think makes great content, and then from there, Claude uses the watch skill to automatically watch it and feed it straight into my second brain. So Claude watches each of these videos, frames, audio, everything, and then comes back with a clean structure and notes about what made the video work.
It fills that straight into the second brain. And this is where things start to compound because the skill and your second brain are watching more and more videos, getting more and more context, and it's getting better and better over time, getting smarter automatically. The second brain side of this whole thing is a video on its own, and I walk through exactly how I run mine, content research, competitor intel, every podcast video I've ever listened to all in one searchable layer in Obsidian.
If that's where you wanna take this, that's the next video to watch. It's linked up here. If this was useful, hit subscribe.
Thanks for watching, and I'll see you in the next one.
The Hook
The bait, then the rug-pull.
Brad opens with a claim that doubles as a threat to every expensive AI video tool on the market: for free, with no proprietary video model, Claude can now watch anything. Before you've hit play, Claude's already an expert on what's in it.
Named ideas worth stealing.
03:43concept
A video is just two things
- Frames
- Transcript
Instead of paying for an expensive multimodal video model, decompose any video into the two things Claude already reads natively — screenshots and timestamped text. Feed both together.
Steal forAny explainer on giving LLMs multimodal context; the "decompose the expensive thing" framing is a teachable pattern
04:40concept
Battle-tested tools, not new wrappers
- yt-dlp (universal video downloader)
- FFmpeg (frame + audio extraction)
Brad explicitly contrasts his use of decade-old, rock-solid CLI tools against MCPs and third-party wrappers. Trust signal: millions of developers, no vendor risk.
Steal forPositioning own-your-stack tools against SaaS wrappers — direct language for MCN+ messaging
05:18model
Frame cap cost scaling
- 1 min -> 60 frames / $0.70
- 10 min -> 80 frames / $0.82
- 30 min -> 100 frames / $0.95
- 1 hr -> 100 frames / $1.62
Capping frames at 100 beyond 30 minutes means cost is nearly flat at scale — a key objection killer for "this will torch my token budget."
Steal forCost math slides for any AI tool pitch; the framing of "surprisingly cheap" deserves its own section in any tutorial
How they asked for the click.
VERBAL ASK
08:25subscribe
“If that's where you wanna take this, that's the next video to watch. It's linked up here. If this was useful, hit subscribe.”
Clean, no hard sell. Next-video link appears visually. Subscribe ask is brief and earned after a dense value delivery.
MENTIONED ON CAMERA
FROM THE DESCRIPTION
OTHER LINKSAlso linked in the description.
Visual structure at a glance.
