Video editing has nowadays become a conversation.. you drop raw footage in a folder, tell Claude Code what you want, and get final.mp4 back. its free, open source, 17,000 stars on Github.. it's called video-use. and the reason it's different from every "AI video editor" is how it actually... works under the hood.. it doesn't guess. it transcribes every word you said first, then edits off the actual transcript. what it does once your footage is in the folder: - cuts every umm, uh, and false start automatically - kills the dead space between takes - color grades each segment.. warm cinematic, neutral punch, or a custom ffmpeg chain - builds the cut with editor sub-agents, runs the animations in parallel - renders a preview, grades its own work, and iterates until it's clean and yeah.. it asks before it cuts. it shows you the plan in plain english and waits for your yes. it's not a black box that spits out a mystery edit.. you stay the director, it does the labor. ▫️ how to start install the skill in claude code (repo below) drop your raw clips in a folder tell it "cut the filler, tighten the pauses, warm cinematic grade" and let it work And yes, it's a tool, not a taste replacement. it'll hand you a clean, tight cut, but the creative calls, the hook, the pacing, the story.. that's still you. it does the 80% that's grunt work so you spend your time on the 20% that matters. editing stopped being a skill you grind for years. it's a prompt now. repo:show more

I built a simpler product. It’s a video editor like iMovie but better. It uses our codex or Claude sub. Also open sourced.

amazing

cool how it works with codex hermes and other agents too

the real win is probably the retry loop underneath, since folder-to-cut only stays "free" if the model doesn't burn cost re-rendering on every failed instruction. Claude Code reads 7 instruction layers before your prompt.

Does it burn ton of tokens because frame analysis / generation is much token heavy than text? Or lib has a way to circumvent it ?

surprisingly not much. i edited a 7 minute for myself it burned 75K tokens

The useful test is whether the agent leaves an editable timeline and a reproducible edit manifest, not just final.mp4. Without that, one bad cut means rerunning a conversation instead of fixing a precise decision.

just dropped hours of editing and got minutes of freedom back

"its a tool not a taste replacement" and "editing stopped being a skill, its a prompt now" are in the same post lol. which one is it. cutting umms is labor, but the edit IS the taste part

Motion design taught me that describing pacing precisely is hard even between two humans. 'Tell Claude what you want' is where this tool will hit the same wall.

claude can do demo video?

Conversation is the interface; reproducibility is the product. I’d want every edit to preserve source hashes, timeline changes, render settings, model/tool versions, and a reversible project file. Then “make it tighter” becomes an auditable edit, not a one-off video artifact.

Does is work in Premiere Pro too? 😇

👍

The best early users may not be broad creators. I’d match this with teams that already have a painful repeatable edit loop: podcast clipping, product demos, course updates, or localization. A narrow workflow cohort will expose the real gaps faster.

I think this may have just transformed my life

This could save creators hours on repetitive editing work

@grok is that 100% true?

The transcript-first workflow finally gives me confidence that cuts aren’t guesswork. I appreciate the explicit confirmation step, it keeps the edit reproducible and safe

if your a real creator shipping tons of content and want your editing to end at uploading raw footage (or want to pay less for an editor that you don't have to train)

this looks like a total game changer honestly

the transcript-first part is the whole unlock. editing off what was actually said instead of guessing at waveforms is why it doesn't butcher the pacing. keeping the human as director is the difference between a tool and a gimmick

Yup most coding agents can do this - explore the world of harnesses and break the duopoly!

this is so useful axel

it's good for simple edits - but these tools still wayyyyy off to do anything really decent. Most are just good at organising, doing the leg work. Ps Prem Pro Ai is pretty dope now tbh

I built a similar product. But with a different angle. Instead of regular edits, you get a high retention optimized cut. It's a video editor that actually understands how to keep your audience watching. you upload raw. choose retention mode. then get ready to post cut in minutes.

Editing off the actual transcript instead of guessing is such a smarter approach.

this is wild just drop the folder, talk to it, get the cut back editing really is just a prompt now 😮

This is a fascinating concept! Using LLMs to automate the tedious parts of video editing, especially transcription-based cuts, could be a huge time-saver for creators. The "you stay the director" aspect is key.

How is it in comparison to hyper frames?

The real unlock here probably isn’t the first cut. It’s keeping the same visual standard across 50 cuts without someone checking every tiny thing.

@grok is it true?