hassancs91/claude-youtube-editor

What it solves

This project automates the tedious process of editing YouTube videos. It allows a creator to record a simple "talking head" video and then uses AI to handle the cutting, visual overlays, audio cleaning, sound effects, thumbnail generation, and uploading, removing the need for traditional video editing software or manual screen recording.

How it works

The pipeline is driven by Claude Code and consists of six primary steps:

1. The Cut: Uses AssemblyAI to transcribe audio and create a clean master cut based on a cuts.json file.
2. The Visuals: Generates UI clones, diagrams, and browser/terminal mockups as code using Remotion (TSX), syncing them to the word-level timing of the transcript.
3. The Voice: Isolates and cleans the voice audio using ElevenLabs or a local RNNoise method.
4. The Sound: Suggests and places sound effects from a library based on the word timing.
5. The Packaging: Generates a title and three A/B test thumbnail options using Gemini.
6. The Upload: Verifies the final video for audio/visual drift and uploads it as a private draft to YouTube via a Python script.

Who it’s for

Content creators and developers who want to produce high-quality, polished YouTube videos with visual overlays and screen-simulated recordings without spending hours in a video editor.

Highlights

• Code-based Visuals: Every screen moment is built as Remotion TSX code rather than being a screen recording.
• Automated Pipeline: A complete end-to-end workflow from raw footage to YouTube upload.
• Brand Customization: Includes a /brand-setup tool to customize fonts, colors, and styles to match a specific channel's identity.
• Integrated AI Stack: Orchestrates multiple AI services including Claude Code, AssemblyAI, ElevenLabs, and Gemini.