The New Viral AI Editor (Turned it Into a Free Skill)
Claude Code turned my raw footage into these 2 edits, and I turned both into free skills. Editor V2 is a fast picture montage learned frame by frame from 1 Instagram reel of 9 seconds. Editor V3 is a cinematic side silhouette learned from a reel by Will Schafer. Here's each edit next to the raw footage, how to install them, the version I threw away and what every version cost 👇 🎁 Both skills, free in my Skool, in the post for this video: https://www.skool.com/nextphase-5790 Stuck? Post it in the Skool and I'll reply, or come to the weekly calls. ⚡ Higgsfield, only for Editor V3 (you connect it to Claude once): https://higgsfield.ai/s/make-your-first-ai-video-yt-jadm-h-NEGQVt 🧰 What you need Claude Code, at least $20 a month. I built both on Opus 5.5 and haven't tested a free model yet. Editing is heavy, so a smaller plan hits its limit faster. No coding: I didn't write a single line. 🛠️ Install (once) 1. Open the Skool post, download the zip and leave it in your Downloads folder. Don't unzip it. 2. Open Claude Code, paste the setup prompt from the Google Drive and hit enter. It finds the zip and installs the free tools (FFmpeg, a speech model, a renderer). If it asks to allow something, click allow. 3. Type "give me a demo of 10 seconds" to see how it looks, then start a new session. 🎬 How to use them Drag in your recording and type 1 line: "Edit this video with Editor V2" (or V3). A few clips of you working help. Before V3 spends a credit, it opens a page in your browser with every shot and what it costs. Give notes with the exact second, and start short. The study tool comes with it, so you can give it any reel you love. ⚙️ Editor V2 About 30 pictures from the reel, used about 75 times, blurred or recolored each time one comes back. Claude cuts my voice, finds footage (Wikimedia Commons, Prelinger Archives), builds the montage on the reel's timing and places every sound. The first montage changed 11 times a second against the reel's 18. Now it's 17. The first full cut took a little over an hour and I didn't touch a clip. 🌑 Editor V3 I recorded twice: front for my voice, side for my outline and mouth. Higgsfield changes the light and background, so it's still my real take. 3 models on the same 5 seconds: • Genjutsu: 7 credits a second, my mouth followed less clearly • Seedance: 7.6, left a piece of my mic in the corner • Kling: 2, kept my mouth moving. That's the one in the skill. Everything after that runs on my Mac. 💥 The bloopers Before these I tried a "1K Editor" and told Claude to use its imagination. A narrator came out of nowhere and my monitor flew at my face. Every time it imagined, it looked like AI slop, so I gave it 1 reel to learn instead. 💸 What it cost • 1K Editor, thrown away: 812 Higgsfield credits. The Editor V3 edit says 1,000. I checked, it's 812. • Editor V3: 217 credits. 127 were my model tests, which you skip. 90 were the final clips for a video a little over 1 minute, about 2 credits per second of silhouette. • Editor V2: 0 new credits. The X-rays of my head are AI images from the 1K round, about 1 credit each, already inside that 812. 📺 My earlier editors Editor V1: https://youtu.be/ge1yzQ5MXmM The free model editor: https://youtu.be/S5Y3Hp2P740 ⏱️ Chapters 0:00 Raw footage vs Editor V2 2:17 Editor V2: learning from 1 reel 4:19 Editor V3: the silhouette edit 7:08 What you need + install 8:42 How to use them 9:44 The bloopers 11:34 What every version cost 12:56 Better than a real editor? 💬 V2 or V3: which one would you post? Tell me in the comments. 🔔 Subscribe for more free skills. See you in the next one. #Claude #ClaudeCode #AIVideoEditing #AIVideoEditor #VideoEditing #ClaudeSkills #ClaudeOpus #Anthropic #Higgsfield #Kling #FFmpeg #SilhouetteEdit #MotionGraphics #ContentCreation #AITools
