F = Motion // Ai explorer.
Joined July 2023
- Tweets71.5K
- Following1.2K
- Followers7.6K
- Likes103K
Force🦅 retweeted
THIS VIDEO COST HIM $0 TO EDIT
No Claude.
No ChatGPT.
No credits.
No editor.
A free open model did the entire edit.
→ Install OpenCode (free agent that reads the same skill folders as Claude Code)
→ Download the Editor V1 skill (built once in Claude, then reused for free)
→ Set the model to GLM 5.3 Flash Free
→ Paste the setup prompt so it installs FFmpeg, Python, and the rest
→ Drag in your raw video and type “edit this video with editor V1”
→ It cuts, adds transitions, graphics, captions, and sound all on your laptop
1.6 million tokens.
$0.00 spent.
No account.
No card.
Bookmark this before you pay for another editing tool.
THIS VIDEO COST HIM $0 TO EDIT
No Claude.
No ChatGPT.
No credits.
No editor.
A free open model did the entire edit.
→ Install OpenCode (free agent that reads the same skill folders as Claude Code)
→ Download the Editor V1 skill (built once in Claude, then reused for free)
→ Set the model to GLM 5.3 Flash Free
→ Paste the setup prompt so it installs FFmpeg, Python, and the rest
→ Drag in your raw video and type “edit this video with editor V1”
→ It cuts, adds transitions, graphics, captions, and sound all on your laptop
1.6 million tokens.
$0.00 spent.
No account.
No card.
Bookmark this before you pay for another editing tool.
TURN CLAUDE INTO A REAL DESIGNER WITH FIVE FREE TOOLS
No more generic AI websites.
→ Taste gives Claude strict design rules so it stops shipping templated, AI looking pages
→ Impeccable upgrades your front end with its own design system and a live browser editor
→ Playwright CLI lets Claude spin up browsers, auto test the site, and take screenshots to verify everything works
→ Awesome Design feeds it real design systems (typography, spacing, buttons, layouts) from actual websites
→ img2threejs turns any reference image into a fully interactive 3D model built entirely in code
All five are free.
Bookmark this so you have the exact stack when you’re ready to ship better sites.
Force🦅 retweeted
THIS IS FCK**** GOLD
MOST PEOPLE STILL DON’T KNOW YOU CAN TEACH CLAUDE TO WATCH YOUTUBE
Drop any video link and it learns the skill itself.
→ Install the free /watch skill from GitHub (`/plugin install watch@claude-video`) so Claude can pull frames + full transcript from any YouTube video
→ Grab a free Gemini API key from Google AI Studio (no card needed) so it can natively read YouTube links
→ Feed it the distill prompt that merges what was seen and what was said into a single timeline of beats
→ It analyzes how the video opens, holds attention, and turns, then pulls the highest-signal lessons with exact timestamps
→ That output becomes a reusable skill your agents can teach themselves and other agents
The people setting this up are already training entire teams of AI agents to pick up new skills overnight.
Bookmark this so you have the exact setup when you’re ready.
THIS IS FCK**** GOLD
MOST PEOPLE STILL DON’T KNOW YOU CAN TEACH CLAUDE TO WATCH YOUTUBE
Drop any video link and it learns the skill itself.
→ Install the free /watch skill from GitHub (`/plugin install watch@claude-video`) so Claude can pull frames + full transcript from any YouTube video
→ Grab a free Gemini API key from Google AI Studio (no card needed) so it can natively read YouTube links
→ Feed it the distill prompt that merges what was seen and what was said into a single timeline of beats
→ It analyzes how the video opens, holds attention, and turns, then pulls the highest-signal lessons with exact timestamps
→ That output becomes a reusable skill your agents can teach themselves and other agents
The people setting this up are already training entire teams of AI agents to pick up new skills overnight.
Bookmark this so you have the exact setup when you’re ready.
THIS IS FCK**** INSANE
MOST PEOPLE THINK THESE AI MOTION GRAPHICS ARE JUST ONE PROMPT
They’re not, It’s a studio you set up once and people are already paying $2,000 $3,000 for the videos it produces.
→ Build the headless render engine with Claude Opus 5.5 (extra high): Node.js + FFmpeg + Playwright
→ Claude writes code that paints every single frame (it doesn’t generate video)
→ Playwright captures every frame in Chrome
→ FFmpeg stitches the frames into the final MP4
→ Feed it real asset references so it stops defaulting to centered all caps text on gradients
→ Ban basic CSS and force closed form spring physics so the motion actually looks good
→ Synthesize the audio in code and lock every cut to the beat
→ Make Claude watch its own frames, critique them, hunt dead frames, and re render until every scene scores 8/10
Set it up once.
Then start selling the output.
Bookmark this for later.
Force🦅 retweeted
THIS GUY HIT $8K/MONTH ON A FACELESS YOUTUBE CHANNEL IN JUST 4 MONTHS
He only spends 3 hours a week. Claude writes every script.
ElevenLabs voices every word. CapCut assembles the rest.
→ Don’t pick a niche you like pick one that pays (finance at $20 RPM can do ~$10k from AdSense alone)
→ Feed Claude the exact niche prompt based on repeatability, audience size, and monetization depth
→ Build the script system so Claude writes hooks with a pattern interrupt in the first 30 seconds, open loops, spoken English, pause markers, visual tags, and a soft CTA
→ Claude also spits out thumbnail concepts and title options (replaces a $200 scriptwriter)
→ ElevenLabs turns the script into voiceover
→ CapCut + Canva assembles the video with stock footage from Pexels
→ Claude writes the title, description, tags, and chapters
→ Check keywords with TubeBuddy or VidIQ
→ Post 1–2 videos the first week, then ramp to 3–4
→ By month 4 add affiliates and sponsorships
600k views at $18 RPM = $10,800 from AdSense alone.
Bookmark this for later.
THIS GUY HIT $8K/MONTH ON A FACELESS YOUTUBE CHANNEL IN JUST 4 MONTHS
He only spends 3 hours a week. Claude writes every script.
ElevenLabs voices every word. CapCut assembles the rest.
→ Don’t pick a niche you like pick one that pays (finance at $20 RPM can do ~$10k from AdSense alone)
→ Feed Claude the exact niche prompt based on repeatability, audience size, and monetization depth
→ Build the script system so Claude writes hooks with a pattern interrupt in the first 30 seconds, open loops, spoken English, pause markers, visual tags, and a soft CTA
→ Claude also spits out thumbnail concepts and title options (replaces a $200 scriptwriter)
→ ElevenLabs turns the script into voiceover
→ CapCut + Canva assembles the video with stock footage from Pexels
→ Claude writes the title, description, tags, and chapters
→ Check keywords with TubeBuddy or VidIQ
→ Post 1–2 videos the first week, then ramp to 3–4
→ By month 4 add affiliates and sponsorships
600k views at $18 RPM = $10,800 from AdSense alone.
Bookmark this for later.
Force🦅 retweeted
Everyone in the closed beta is already in the draw
The only filter left is whether they noticed you
Well i am sat for what @perpdotcom is cooking
We’re giving away an iPhone Duo 📱
Everyone selected for the Perp.com closed beta is in the draw.
Get seen. Get in.
Force🦅 retweeted
THIS OPEN SOURCE REPO JUST GAVE EVERYONE CLAUDE MOTION DESIGN FOR FREE
You don’t need the Teams or Enterprise plan. It creates commercial grade product launch films with pure code instead of paid video AI.
→ Search “Huashu Motion GitHub” (it’s the first result)
→ Use it as the open source version of Claude Motion
→ Generate all motion graphics with code (JavaScript and similar)
→ It adds transitions, music, and a full browser-based editor
→ Point at frames, type changes, and it rebuilds the exact spot
→ Compare it head to head with other tools on product demos
No Kling.
No Seedance.
Just code that outputs finished films.
Bookmark this for later.
THIS OPEN SOURCE REPO JUST GAVE EVERYONE CLAUDE MOTION DESIGN FOR FREE
You don’t need the Teams or Enterprise plan. It creates commercial grade product launch films with pure code instead of paid video AI.
→ Search “Huashu Motion GitHub” (it’s the first result)
→ Use it as the open source version of Claude Motion
→ Generate all motion graphics with code (JavaScript and similar)
→ It adds transitions, music, and a full browser-based editor
→ Point at frames, type changes, and it rebuilds the exact spot
→ Compare it head to head with other tools on product demos
No Kling.
No Seedance.
Just code that outputs finished films.
Bookmark this for later.
Force🦅 retweeted
OPUS 5.5 JUST BUILT FULL ANIMATED 3D WEBSITES FROM A FEW PINTEREST REFERENCES
It is very cheap.
He never ran out of credits on the cheapest Claude plan while building the whole thing.
→ Go to Motion Sites and copy a prompt/template
→ Open Claude on Opus 5.5 and build it out
→ Connect the Higgsfield connector
→ Find reference images or videos on Pinterest and drag them in
→ Prompt Claude to generate matching assets with Higgsfield (GPT Image 2.5) and replace them on the site
→ Iterate with simple follow-up prompts for the rest of the images, coin, and sections
→ For from scratch builds: new folder, drop a reference video + image, and tell it to recreate the exact animation and UI (using Three.js or whatever it needs)
→ Preview, then drag the folder to Vercel to publish
One prompt turns references into a fully animated, responsive website.
Bookmark this for later.
OPUS 5.5 JUST BUILT FULL ANIMATED 3D WEBSITES FROM A FEW PINTEREST REFERENCES
It is very cheap.
He never ran out of credits on the cheapest Claude plan while building the whole thing.
→ Go to Motion Sites and copy a prompt/template
→ Open Claude on Opus 5.5 and build it out
→ Connect the Higgsfield connector
→ Find reference images or videos on Pinterest and drag them in
→ Prompt Claude to generate matching assets with Higgsfield (GPT Image 2.5) and replace them on the site
→ Iterate with simple follow-up prompts for the rest of the images, coin, and sections
→ For from scratch builds: new folder, drop a reference video + image, and tell it to recreate the exact animation and UI (using Three.js or whatever it needs)
→ Preview, then drag the folder to Vercel to publish
One prompt turns references into a fully animated, responsive website.
Bookmark this for later.
Force🦅 retweeted
CLAUDE CODE CAN NOW TURN A FEW REFERENCE IMAGES INTO A FINISHED ANIMATED EXPLAINER
Almost nobody is using this yet, and the whole thing takes under five minutes.
→ Grab a few images of the look you want (movie still or magazine ad) and save them
→ Open Claude Code on Opus 5.5 with thinking set to high and drag in your references
→ Install the animation skill
→ Type in your topic and format
→ Claude storyboards every scene and double checks its own work
→ It stitches everything into one clean MP4
Full breakdown is in the video.
Bookmark this for later.
CLAUDE CODE CAN NOW TURN A FEW REFERENCE IMAGES INTO A FINISHED ANIMATED EXPLAINER
Almost nobody is using this yet, and the whole thing takes under five minutes.
→ Grab a few images of the look you want (movie still or magazine ad) and save them
→ Open Claude Code on Opus 5.5 with thinking set to high and drag in your references
→ Install the animation skill
→ Type in your topic and format
→ Claude storyboards every scene and double checks its own work
→ It stitches everything into one clean MP4
Full breakdown is in the video.
Bookmark this for later.
Force🦅 retweeted
THE #1 TRENDING GITHUB REPO RIGHT NOW CAN REVERSE ENGINEER ANY APP
It’s called Reverse Engineer Anything, and it already has tens of thousands of stars.
→ Point your coding agent at any app (or even old hardware)
→ It inspects the software and breaks down how the feature actually works
→ It traces the logic behind it instead of you guessing
→ Then it builds your own version for your project
Not just apps people are also using it to revive old hardware like keypads, scanners, and sound cards.
Full breakdown is in the video.
Bookmark this for later.
THE #1 TRENDING GITHUB REPO RIGHT NOW CAN REVERSE ENGINEER ANY APP
It’s called Reverse Engineer Anything, and it already has tens of thousands of stars.
→ Point your coding agent at any app (or even old hardware)
→ It inspects the software and breaks down how the feature actually works
→ It traces the logic behind it instead of you guessing
→ Then it builds your own version for your project
Not just apps people are also using it to revive old hardware like keypads, scanners, and sound cards.
Full breakdown is in the video.
Bookmark this for later.
Force🦅 retweeted
THE BIGGEST SKILL IN 2026 IS HOW TO MAKE AI ADS
This full workflow turns one product photo into a consistent AI UGC video.
→ Find and download an image of the product
→ Upload it to ChatGPT and create a product sheet (keeps the product consistent across angles and text)
→ In the same chat, ask for 3 AI text prompts for avatars that can promote the product
→ Pick one and say “create the character sheet for avatar number one” (multiple angles, expressions, and holding the product)
→ Ask for 3 AI UGC video ideas for a 15 second video based on the avatar and product
→ Choose or refine one idea, then say “give me the AI video text prompt”
→ Copy the video prompt + character image + product image
→ Paste them into an AI video model (text prompt first), set duration/aspect ratio, and generate
Same chat the whole way so the product and avatar stay consistent.
Iterate the prompt or generate again if the first result isn’t perfect.
Full breakdown is in the video.
Bookmark this for later.
THE BIGGEST SKILL IN 2026 IS HOW TO MAKE AI ADS
This full workflow turns one product photo into a consistent AI UGC video.
→ Find and download an image of the product
→ Upload it to ChatGPT and create a product sheet (keeps the product consistent across angles and text)
→ In the same chat, ask for 3 AI text prompts for avatars that can promote the product
→ Pick one and say “create the character sheet for avatar number one” (multiple angles, expressions, and holding the product)
→ Ask for 3 AI UGC video ideas for a 15 second video based on the avatar and product
→ Choose or refine one idea, then say “give me the AI video text prompt”
→ Copy the video prompt + character image + product image
→ Paste them into an AI video model (text prompt first), set duration/aspect ratio, and generate
Same chat the whole way so the product and avatar stay consistent.
Iterate the prompt or generate again if the first result isn’t perfect.
Full breakdown is in the video.
Bookmark this for later.