I’ve been sitting on a video idea for months. Nothing fancy—just a walkthrough of a side project I built. Every time I thought about actually making it, I’d remember how much I hate video editing. The timeline, the keyframes, the rendering. It’s not that I can’t do it. It’s that every minute spent tweaking bezier curves feels like a minute stolen from actually building things.
Then I tried Remotion’s Agent Skills with Claude Code, and something clicked.
What’s Actually Happening Here
Remotion has been around for a while. It’s a React framework for programmatic video creation—think of it like building a webpage that renders to video instead of a browser. What’s new is that they released Agent Skills, which is basically a knowledge pack that lets Claude Code understand Remotion’s architecture, APIs, and patterns.
The way it works in practice: you describe what you want in plain English, and Claude generates the actual Remotion code. Not just snippets—full, working compositions that you can render.
What I find interesting about this isn’t the “AI generates video” angle. It’s how it changes the relationship between having an idea and seeing it exist. The bottleneck shifts from execution to clarity of thought. If you can clearly describe what you want, you can have it.
Why This Feels Different
I’ve played with other AI video tools. Most of them are black boxes—you feed it prompts and hope something usable comes out. Remotion Agent Skills sits in a different spot on the abstraction ladder. It’s generating code, not pixels. That means you can inspect what it produces, tweak it, understand it.
There’s something deeply satisfying about this approach. It reminds me of when I first discovered that I could automate boring tasks with scripts. The difference is that instead of automating file operations or data processing, you’re automating the fiddly parts of visual storytelling.
Let me give you a concrete example. I wanted a simple sequence: code appears on screen, line by line, then the camera pans to a diagram, then some text fades in. Pre-AI, I’d spend an hour in After Effects or a similar tool. With Remotion Agent Skills, I typed something like “show this code appearing line by line over 2 seconds, then pan to this SVG diagram at 3 seconds, then fade in ‘x: 100, y: 200’ at 5 seconds.” Thirty seconds later, I had a working Remotion composition.
The point isn’t that this is revolutionary. The point is that it removes enough friction that you might actually make the thing you’ve been thinking about making.
The Technical Reality
Under the hood, Remotion is clever. It treats video as a rendering problem, not a editing problem. Your video is just a React component that gets called for every frame. This means you can use all the React tooling you already know—component composition, state management, conditional rendering. The Agent Skills know how to map natural language to these primitives.
What the Skills are doing, essentially, is codifying patterns. When you say “smooth fade in,” it knows to write <Fade in={true} durationInFrames={30} />. When you say “pan from left to right,” it knows to use interpolate() and spring() animations.
This is where I think it gets interesting for indie creators and solo developers. You’re not locked into some vendor’s idea of what video editing should look like. You’re working with code. If the AI generates something that’s almost right but not quite, you can just open the file and tweak it. You can build your own reusable components. You can version control your videos.
I keep thinking about how this changes the math for side projects. If you’re building documentation, or explaining a concept, or sharing something you made, video is often the best medium. But the cost of entry has been so high that most of us just don’t bother. What becomes possible when video is roughly as much work as writing a blog post?
What I’ve Actually Made
I don’t want to oversell this. It’s not magic. You still need to have taste and make decisions. But here’s what I’ve managed to create in the past few weeks:
A video explaining how a particularly gnarly piece of my code works. I dropped in the code file, asked for a walkthrough with callouts highlighting specific functions, and got back something I could actually use.
A quick update video for a small app I maintain. Showed the old UI, transitioned to the new UI, highlighted the changes. Maybe 3 minutes of work.
Some social media clips promoting a blog post. Nothing fancy—just the title, some key points as text, and a call to action. But I made five variations in the time it would’ve taken me to make one the old way.
The common thread: none of these were important enough to justify spending hours in traditional video software. But they were useful to have. They’re the sorts of things that live in the gap between “not worth the effort” and “nice to have.”
The Limits
I should be honest about what this isn’t good for yet.
Complex narratives still require a lot of hand-holding. If you’re trying to tell a story with specific emotional beats, you’ll spend a lot of time iterating. The AI can help with execution, but you still need to direct.
Anything that requires precise timing—a music video, say, or something that needs to hit specific beats—will still need manual adjustment. The AI can get you close, but “close” isn’t enough when you’re cutting to a drop.
And there’s still a learning curve. You get better results if you understand Remotion’s mental model. The Agent Skills are force multipliers, not replacements for understanding.
Why This Matters for Indie Creators
Here’s what I keep coming back to: we’re living through a strange moment where the tools for creation have been democratized, but the tools for presentation haven’t quite caught up. Anyone can build an app, write a blog, record a podcast. But making your work stand out—making it polished and professional—still often requires specialized skills or budget.
Remotion Agent Skills feels like a step toward closing that gap. It doesn’t make you a great video creator. But it might make you a video creator at all, where before you weren’t.
What I’m noticing is that video creation is going through something similar to what happened with coding. The technical execution is becoming a commodity. What matters is judgment, taste, and having something worth saying.
There’s something poetic about this. The skills that differentiate you—knowing your domain deeply, having a unique perspective, caring about the details—are exactly the skills that become more valuable when the rote work is automated.
Learning More
If you’re curious to try this yourself, here’s what I’d recommend:
Start with the official Remotion Skills documentation. It’s straightforward and assumes you know basic React.
Watch this tutorial to see what the workflow looks like in practice. Sometimes seeing someone else do it makes it click.
And if you want to understand the philosophy behind Remotion (not just the AI stuff), this deep dive from Jonny Burger, the creator, is worth your time.
The Part That’s Hard to Quantify
I’ve focused on the practical stuff—how it works, what you can make, the technical details. But there’s a softer benefit I’m not sure I’ve fully captured.
There’s a joy in having an idea and immediately acting on it. In not having to context switch between “creator” and “video editor.” In the flow state that comes when your tools keep up with your imagination.
I made a video last week just because I wanted to see how something would look. Not for work, not for an audience. Just for me. That’s not something I would have done before. That’s the change I’m most excited about.
Video creation stopped being a chore. And I think that matters.
This article reflects my early experiments with Remotion Agent Skills. I’m sure I’ll discover more use cases—and more limitations—as I spend more time with it. If you try it and make something interesting, I’d love to see what you build.