01
Search what was said
Use multiple terms or quoted phrases across timestamped captions. Every result is a direct jump back into context.
Open source · Local first · Built for editors
Find the line.
Keep the moment.
Cut faster.
ClipSift turns spoken words and finished scripts into exact, review-ready clip maps—without forcing your production library into someone else’s workspace.
- 01 Search transcripts
- 02 Build clip maps
- 03 Render locally
ClipSift youtube.com/watch?v=…
LOCAL
One focused workspace
React + TypeScriptMIT licensedNo analyticsProject data stored locally
The useful parts, together
Stop hunting through footage.
ClipSift keeps the video, transcript, decisions, and output queue close enough that you never lose the thread.
02
Shape clean moments
Save in/out ranges, nudge edit points, add notes, and move each moment from candidate to selected to used.
03
Start from a script
Break narration into visual beats, discover candidate footage, and review the whole story in script order.
04
Render where you work
Queue local MP4 cuts with yt-dlp, watch job progress, retry failures, and optionally send finished files through rclone.
05
Hand off a real edit plan
Export CSV, JSON, Markdown, plain text, or Premiere marker CSV. Your decisions leave the app in useful formats.
CSVJSONMDTXTPREMIERE
A shorter path to the cut
From source to select.
Work a single recording in detail, or use a finished script to coordinate sources across a larger story.
-
01
Load the source
Paste a YouTube URL, pull its transcript, or import SRT, VTT, and plain text captions.
Video + transcript -
02
Find and decide
Search exact language, save ranges, compare candidates, and record the choices that matter.
Search + review -
03
Render or hand off
Queue MP4s locally or export a structured clip map for the editor and timeline.
Clips + edit plan
Mode 01
One video.
Every useful line.
For interviews, press conferences, podcasts, and game film. Search the full transcript, save the strongest moments, and export only what belongs in the cut.
- Timestamped transcript
- Editable in/out points
- Notes, labels, and statuses
Mode 02
One script.
A complete clip map.
For essays, profiles, documentaries, and social packages. Keep source discovery and visual coverage aligned to the actual narration.
- Ordered story beats
- Ranked source candidates
- Review and render queue
Open source by design
Your workflow should stay yours.
ClipSift is MIT licensed. Read the code, run it locally, adapt it to your production stack, and keep control of the database and rendered files.
Maintained by Art Moreno · MIT License
$ git clone https://github.com/ArtMoreno/clipsift-web.git
$ cd clipsift-web
$ npm ci
$ npm run dev
Keep independent tools moving
ClipSift saved you a search?
Support Art’s open-source work with a coffee.