SkillsMar 29, 2026·2 min read

Remotion Rule: Extract Frames

Remotion skill rule: Extract frames from videos at specific timestamps using Mediabunny. Part of the official Remotion Agent Skill for programmatic video in React.

TL;DR
This Remotion skill rule teaches AI agents to extract frames from videos at specific timestamps using the Mediabunny library in React.
§01

What it is

This is a Remotion skill rule that teaches AI coding agents how to extract frames from videos at specific timestamps using Mediabunny. It is part of the official Remotion Agent Skill for programmatic video creation in React.

The rule targets developers using Remotion for video generation who need to extract thumbnails, create filmstrips, or process individual frames from video files. It works with Claude Code, OpenAI Codex, and Cursor.

§02

How it saves time or tokens

The rule activates automatically when the agent detects frame extraction work in a Remotion project. It provides the exact function signatures and usage patterns for Mediabunny's extractFrames API, eliminating the need to look up documentation or experiment with parameters.

With an estimated 500 tokens per activation, the rule adds minimal overhead to the agent's context while providing precise, actionable guidance.

§03

How to use

  1. Install the Remotion skills collection with npx skills add remotion-dev/skills.
  2. Open your Remotion project and ask your coding agent to extract frames from a video.
  3. The rule activates automatically and the agent generates the correct Mediabunny code.
§04

Example

import { extractFrames } from '@mediabunny/core';

const frames = await extractFrames({
  src: 'https://example.com/video.mp4',
  timestamps: [0, 5, 10, 15],
  format: 'png',
  width: 1920,
  height: 1080,
});

for (const frame of frames) {
  console.log(`Frame at ${frame.timestamp}s: ${frame.url}`);
}
§05

Related on TokRepo

§06

Common pitfalls

  • The extractFrames function requires a valid video URL accessible from the server. Local file paths must be served via a local HTTP server or use file:// protocol.
  • Extracting many frames from a long video can be memory-intensive. Process frames in batches rather than requesting all timestamps at once.
  • The output format (png vs jpg) affects file size and quality. Use png for lossless quality when compositing, jpg for smaller file sizes in thumbnails.

Frequently Asked Questions

What is Mediabunny?+

Mediabunny is a media processing library used within the Remotion ecosystem for video manipulation tasks like frame extraction, transcoding, and thumbnail generation. It provides programmatic access to video processing functions that Remotion skills can leverage.

Does this rule work without Remotion?+

The rule is designed for Remotion projects and activates in that context. The Mediabunny extractFrames function itself can work independently, but the skill integration and automatic activation require a Remotion project setup.

How many frames can I extract at once?+

There is no hard limit on the number of timestamps, but memory usage scales with frame count and resolution. For videos with many frames, process in batches of 10-20 timestamps to avoid memory pressure.

What video formats are supported?+

Mediabunny supports common video formats including MP4, WebM, and MOV. The underlying decoder handles most codecs supported by FFmpeg. Ensure your video file is accessible and properly encoded.

Can I customize the output resolution?+

Yes. The extractFrames function accepts width and height parameters. You can extract frames at any resolution, independent of the source video resolution. Omitting these parameters uses the source video dimensions.

Citations (3)
🙏

Source & Thanks

Created by Remotion. Licensed under MIT. remotion-dev/skills — Rule: extract-frames

Part of the Remotion AI Skill collection on TokRepo.

Discussion

Sign in to join the discussion.
No comments yet. Be the first to share your thoughts.