Free 'video-use' editor emerges

A free open-source tool called video-use surfaced on GitHub this month, promising to turn raw footage into a finished video by chatting with Claude Code. (github.com, anthropic.com) The repository, published by the browser-use team, had about 2,800 GitHub stars and 381 forks when checked on April 23, 2026. Its README says users can drop clips into a folder, ask Claude Code to edit them, and get back an `edit/final.mp4`. (github.com, github.com) Video-use says it cuts filler words and dead space, adds 30-millisecond audio fades at cuts, burns in subtitles, auto-grades color, and generates animation overlays with tools including Manim and Remotion. The project says it stores session memory in a `project.md` file so later edits can pick up where an earlier session stopped. (github.com) The basic idea is transcript-first editing: instead of scrubbing a timeline by hand, the model reads word-level timestamps and uses those text markers to decide where to cut. The README says the system relies on an ElevenLabs Scribe transcription pass for word timing, speaker labels, and audio-event tags such as laughter or applause. (github.com, elevenlabs.io) That puts video-use in the same workflow lane as products that already sell AI-assisted cleanup and captioning. Descript’s pricing page lists tools including Remove Filler Words and Studio Sound on paid plans starting at $16 a month, while Microsoft Clipchamp advertises AI silence removal, subtitles, color correction, and noise suppression in its browser editor. (descript.com, clipchamp.com) Adobe still sells Premiere as part of a subscription business, either inside the broader Creative Cloud bundle or as a standalone Premiere plan. CapCut also now steers users toward paid memberships, with its April 8, 2026 help page describing an updated subscription structure for new subscribers and a separate teams page saying 2026 pricing typically starts around $15 to $30 per seat each month when billed annually. (adobe.com, adobe.com, capcut.com, capcut.com) Free editors are not new. OpenShot has been around since 2008 as a free open-source editor for Linux, Mac, and Windows, but its model is a traditional timeline with manual effects, titles, and trimming rather than a chat-driven agent that proposes edits from transcripts. (openshot.org, github.com) Video-use also is not fully free in every part of the stack. The setup instructions call for `ffmpeg`, optionally `yt-dlp`, and an ElevenLabs API key, and ElevenLabs’ current pricing pages show speech and audio services are usage-metered rather than unlimited. (github.com, elevenlabs.io, elevenlabs.io) Forks are already pushing the idea in different directions. One recent fork, `video-use-premiere`, swaps in local Whisper transcription, adds visual and ambient-sound analysis, and exports cut files for Premiere Pro, DaVinci Resolve, and Final Cut Pro instead of only a finished MP4. (github.com) What happens next is the same question hanging over a lot of AI production software in 2026: whether a GitHub demo can hold up on messy real projects. For now, video-use is a public repo with a fast-growing audience, a transcript-driven editing pitch, and a direct shot at software people currently rent by the month. (github.com, descript.com, adobe.com)