assemble
Skill16 starsChanged 4 months agoMITView on GitHub
Cut a beat's segment out of the source, and drop the generated asset back into the cut — with ffmpeg (trim, audio re-lay, overlay, concat). Use to extract src.mp4 before generation, and to place the approved asset after GATE 2.
2 other projects use a version of this.
Adopted by
Most recently taken up by
- lucasacoutinho/ddstheir own version, first seen Oct 13, 2026
- etinpres/assembletheir own version, first seen May 6, 2026
Other versions2 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 2
- etinpres/assemble★ 2same name
- lucasacoutinho/dds★ 2same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository5 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- analyze-videoSkill
- craft-promptSkill
- graphic-designSkill
- omni-vfxSkill
The file
We split this skill into the rules it gives an agent
3 rules. About 0 are common, 1 is distinctive, and 2 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈1
Distinctive · estimate
This author's own choices. Worth a look.
2
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 30
Use out-voiced.mp4 as the beat's asset.
- Very specific: line 46
If the segments differ in codec/size and concat stutters, re-encode instead of -c copy, or use the xfade path below for a clean …
- Distinctive: line 65
Use this for the "editing effect" sweep and for smoothing the seams where a transformed segment meets the untouched footage.
≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
assemble — cut + place (ffmpeg)
Paths —
input/,analysis/,beats/,assets/,output/are relative to the active projectprojects/<slug>/(the video being edited). Per-beat video work lives inbeats/<seg>/; generated graphics live inassets/refs/.
Two jobs, one skill: cut the segment before generating, place the asset after the user approves. All of it is plain ffmpeg.
Cut (before generation) — to an EXACT even-second bucket (the sync rule)
Extract the beat window into beats/<seg>/src.mp4 at an EXACT 4 / 6 / 8 / 10 s length (pick the smallest bucket that fully covers the moment; snap the in-point so the whole action is inside). Cut by duration (-t <bucket>), not by an out-point, so the length lands exactly on the bucket. Omni is then submitted with duration = that same bucket, and returns the output at the same length → sync holds natively, no time-lock needed (see omni-vfx). KEEP the audio (-c:a aac, never -an) — we re-lay it after Omni:
ffmpeg -ss <start_sec> -i input/<clip>.mp4 -t <bucket> # <bucket> ∈ 4 | 6 | 8 | 10
-c:v libx264 -r 30 -g 30 -keyint_min 30 -pix_fmt yuv420p -c:a aac -movflags +faststart \
beats/<seg>/src.mp4Confirm it landed exactly on the bucket: ffprobe -v error -show_entries format=duration -of default=nk=1:nw=1 beats/<seg>/src.mp4. Record the true source in-point (<start_sec>) — the transformed segment is placed back at that in-point at assembly.
Dense keyframes (-g 30) matter — sparse keyframes make the face freeze on seek downstream.
Also keep the original audio around separately —
ffmpeg -ss <start> -to <end> -i input/<clip>.mp4 -vn -c:a aac beats/<seg>/orig-audio.m4a— you'll mux it back over the Omni output (next).
Restore the real voice (right after an Omni clip lands) — REQUIRED for vfx
Omni regenerates the audio track — it will replace the speaker's voice with a synthetic one. Never ship Omni's audio. Because Omni preserves timing, the original audio lines up exactly, so re-lay it:
ffmpeg -i beats/<seg>/out.mp4 -i beats/<seg>/orig-audio.m4a \
-map 0:v:0 -map 1:a:0 -c:v copy -c:a aac -shortest beats/<seg>/out-voiced.mp4Use out-voiced.mp4 as the beat's asset. (If the source window truly had no speech, Omni's audio can be dropped entirely instead.)
Place (after GATE 2)
Drop the approved asset back onto the timeline. Two cases:
A. vfx (out-voiced.mp4) — replaces the original window
The transformed segment takes the same in/out. Split the source around the window and concat:
# head: 0 → start, tail: end → duration (scratch goes in beats/_scratch/, which is gitignored)
mkdir -p beats/_scratch
ffmpeg -i input/<clip>.mp4 -to <start> -c copy beats/_scratch/head.mp4
ffmpeg -i input/<clip>.mp4 -ss <end> -c copy beats/_scratch/tail.mp4
printf "file '%s'\nfile '%s'\nfile '%s'\n" \
"$PWD/beats/_scratch/head.mp4" "$PWD/beats/<seg>/out-voiced.mp4" "$PWD/beats/_scratch/tail.mp4" > beats/_scratch/list.txt
ffmpeg -f concat -safe 0 -i beats/_scratch/list.txt -c copy output/<clip>-final.mp4If the segments differ in codec/size and concat stutters, re-encode instead of -c copy, or use the xfade path below for a clean cross-dissolve at the seams.
B. graphic (assets/refs/<name>.png) — overlays on top of the playing video for the beat's window
The graphic-design output for this beat (a transparent PNG in assets/refs/) is composited over the base for [start,end]. Add a quick fade so it doesn't pop:
ffmpeg -i input/<clip>.mp4 -i assets/refs/<name>.png -filter_complex \
"[1]format=rgba,fade=in:st=<start>:d=0.3:alpha=1,fade=out:st=<end-0.3>:d=0.3:alpha=1[g]; \
[0][g]overlay=enable='between(t,<start>,<end>)':x=(W-w)/2:y=(H-h)/2" \
-c:a copy output/<clip>-final.mp4- Corner instead of centered: set
x=W-w-40:y=H-h-40. - Scale the graphic first if it's oversized: add
scale=iw*0.6:-1in the[1]chain.
Transitions & sweeps (beat 1b and any seam) — ffmpeg xfade
ffmpeg ships ~50 transitions (fade, wipeleft, slideup, zoomin, dissolve, circleopen…). Cross between two clips:
ffmpeg -i A.mp4 -i B.mp4 -filter_complex \
"[0][1]xfade=transition=zoomin:duration=0.5:offset=<A_dur-0.5>,format=yuv420p" out.mp4Use this for the "editing effect" sweep and for smoothing the seams where a transformed segment meets the untouched footage.
Stitch several transformed segments
Lay each transformed segment back at its real in-point (case A per beat), left to right, working through the beats in order — or, if many, re-encode all pieces to one common format and concat, then overlay the graphics (case B) last. xfade at the joins keeps cuts from jarring.
Notes
- Can't matte cleanly with ffmpeg — you can't key the subject out to put a graphic behind them. Options: composite the graphic in front (case B), or add a kie background-removal step if "behind" is essential.
- Animated overlays (bouncy slide-ins, count-ups) are painful in pure ffmpeg. Prefer generating motion inside Omni, or use a simple fade (case B). Fancy kinetic graphics are out of scope for the ffmpeg-only kit.
- Render straight to
output/— ffmpeg streams, so no temp-frame blowup.
Status
On place, set the beat → ✅ placed in beat-plan.md. When all beats are placed, the final is in output/.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

