video-editing-craft
Skill0 starsChanged 16 days agoMITView on GitHub
How to edit a video so it holds attention - pacing and tempo design, cuts per minute, when to cut fast or slow, A-roll vs B-roll vs graphics rhythm, punch-in zooms and highlights, invisible continuity, the cut vocabulary (jump, J, L, match, smash, cutaway) and editing habits. Use when the user wants a video edited, tightened, made punchier or more engaging, asks why an edit feels flat or slow, wants a rough cut reviewed, or asks how to pace, cut or structure a video for YouTube, Shorts, Reels or TikTok.
In this repository13 other files it gives its agents.
- 2bft-video-editorSkill
- asset-librarySkill
- comedy-editing-timingSkill
- editor-osSkill
- edit-videoSkill
- ffmpeg-edit-toolkitSkill
- local-av-toolsSkill
- long-to-shortsSkill
- motion-graphics-for-videoSkill
- quest-lab-editorSkill
- talking-head-polishSkill
- transcribe-localSkill
- video-sound-and-musicSkill
The file
We split this skill into the rules it gives an agent
4 rules. About 0 are common, 4 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈4
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 12
Every edit choice serves one experience, and a choice that fights it loses viewers even if it is "better editing".
- Distinctive: line 20
Ask which one if it is not obvious.
- Distinctive: line 37
Switch what is on screen every few seconds: when the goal is engagement.
- Distinctive: line 40
Captions: only to stress words or fill genuinely empty visual space.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 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
Video editing craft
Learned from tutorials by working editors (sources at the bottom) and turned into decisions you can apply to a real timeline. Use with ffmpeg-edit-toolkit to carry the decisions out, and with comedy-editing-timing, video-sound-and-music and motion-graphics-for-video for those layers.
1. Decide the experience before any cut
Every edit choice serves one experience, and a choice that fights it loses viewers even if it is "better editing".
| The viewer came for | Edit like this | Avoid |
| Hanging out with someone (vlog, talk, podcast cut) | Remove dead parts only; long takes are fine (one example channel averaged about 90 seconds between cuts); keep it authentic | Flashy animation, rapid cuts that break the intimacy |
| Entertainment (challenge, comedy, explainer, shorts) | Remove every pause; change what is on screen every few seconds; sound and graphics on every beat | Long static A-roll |
| Learning something | Clear B-roll or graphics for each claim, pace set by the explanation, calm music | Visual noise that competes with the lesson |
Ask which one if it is not obvious. When unsure, favour what the footage can honestly support.
2. Design tempo, don't let it happen
- Variation is the product. A video at one constant speed feels either relentless or dull. Plan fast and slow stretches like a roller coaster: where is the slow climb, where is the drop.
- Start from the emotion, then pick the pace. Reflective, tender or heavy moment: cut slower so it lands. Tense, rushed or exciting: cut faster. Write the emotional destination of each section first.
- Measure it. Cuts per minute (CPM) is the editor's speedometer: count the cuts in a 30-60 s sample and scale. One worked example went from about 10 CPM to about 33 CPM when a chase began. Change tempo at story change points: more energy needed, raise CPM; more introspection, lower it. Never change it at random, because random pace changes read as amateur.
- Ask five questions per scene: does it have one overall tempo, what is it, does it change, where, and why.
- Four things set perceived pace, and they add up: how fast the cuts are; how much movement is in the frame; how close the subject is (tight and busy feels faster than wide); how much the camera moves. Lots of cuts plus whip pans plus close-ups is much faster than cuts alone. To slow a scene, change any of them, not just the cut count.
- Exercise: cut a scene at the intended tempo, duplicate it, and cut the opposite tempo. Compare. Then try changing tempo mid-scene for a reason.
- Speed contrast works inside a clip too: ramp 400-800% fast sections against normal speed. Film at 60 fps if you may want 50% slow motion that still looks smooth.
- Short-form is simple and fast; long-form is varied and complex. Don't paste short-form tempo onto a 20-minute video.
3. Keep the picture changing, without turning to mush
- A-roll (you see and hear the speaker) feels personal and holds attention; use it for short stretches and the most important statements, spoken with confidence.
- B-roll shows what the words describe, explains faster than a face does, and is more interesting than A-roll: use as much as the story supports. Stock footage is the lazy version; motion graphics are the strongest B-roll for anything abstract or complicated.
- Switch what is on screen every few seconds when the goal is engagement. But do not shorten shots just to shorten them: a 10-second shot is fine if it is interesting for 10 seconds.
- Add life without a new clip: slow scale or position drift on stills; an abrupt punch-in on the line that matters (see the zoom recipe); a tiny shake only if it suits the style.
- Guide the eye in an image or screen: animate or underline the key text, darken or blur the surroundings, circle or arrow it, tint it (red reads as negative, yellow or green as positive), make the subject glow.
- Captions only to stress words or fill genuinely empty visual space; three words or fewer at a time; don't use them as filler.
- Voiceover beats on-camera explaining for facts: one example condensed a long explanation to 23 seconds of voiceover; it felt like information, not a lecture. Keep sections of a multi-part video to a few minutes each.
4. Make the edit invisible, then break it on purpose
- Where the eye is at the end of shot A should be where it is at the start of shot B. Cutting from eyes at one side of the frame to a subject at the other makes the viewer refocus and breaks immersion.
- A graphic should not just appear. Move it in, or cover the arrival with a sound effect (pop, shutter, whoosh).
- Use a full-screen transition to hide a cut when shots can't be matched.
- Deliberately break continuity now and then (a smash cut, an abrupt music stop): the jolt wakes attention. It only works against a smooth baseline.
- Transitions: simple beats flashy. Reserve effects for moments where they mean something.
5. Habits that save the edit
- Finish each section completely (cut, graphics, sound, colour) before starting the next; "70% of everything, then a second pass" costs more time.
- Perfection doesn't exist; from about 90% to 100% is ego. Details only you and a few peers will notice are permitted.
- Trim the dead start: viewers decide within the first seconds. Open on the strongest moment or the question.
- End on a high, not a fade-out of energy.
- Watch for a flat edit: usually it has no tempo change, no sound design, and the same kind of shot too long.
Review checklist
- Experience chosen and every choice serves it
- Emotional destination per section; tempo matches it; tempo changes have reasons
- CPM measured before and after (state both)
- Nothing on screen unchanged for longer than it earns
- Eye-trace across cuts; graphics enter with motion or sound
- First 5 seconds earn the next 30; ending lands
- Sound pass done (see
video-sound-and-music)
More
- Cut-by-cut guide with when to use each:
references/cut-vocabulary.md - Comedy:
comedy-editing-timing. Sound and music:video-sound-and-music. Graphics:motion-graphics-for-video. Commands:ffmpeg-edit-toolkit. Where to get music, SFX, memes, B-roll, fonts:asset-library. Clean-up (noise, stems, cutouts, shot detection):local-av-tools.
Sources (paraphrased, not quoted)
- Inside The Edit, "Pacing in Film Editing: Techniques Every High-End Editor Must Know", https://www.youtube.com/watch?v=CdyTIng0M8Q
- Austen Menges, "How PRO Video Editors Change RHYTHM For EMOTION", https://www.youtube.com/watch?v=jY6hT8-SR3g
- Learn By Leo, "How to edit SO good your viewers get addicted to your videos", https://www.youtube.com/watch?v=sLgHqZSe2o0
- Digital Spaghetti, "Doobydobap's Editing Hacks to Improve Your Pacing", https://www.youtube.com/watch?v=fH_6iL2EmHM
- Shutterstock Tutorials, "9 Cuts Every Video Editor Should Know", https://www.youtube.com/watch?v=Wv3Hmf2Dxlo
- Jonny Duncan, "10 Editing Cuts Every Video Editor Should Know", https://www.youtube.com/watch?v=YtULPT1aBWM
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.

