Skip to content
Back to skills

Charles Claude Editing

ASecurity

Edit raw talking-head videos into engaging, organic-looking Instagram Reels / TikToks - cut dead space and stumbles, pick the best take, 2-3 word captions, retention-maxing image/video overlays (pop-culture stickers, cutouts, split screens, tweet screenshots), sound effects, a background track from the music folder, and a TikTok-title-style hook when asked. Covers news and education/entertainment videos. Use this whenever Charles shares a talking-head .MOV/.mp4 and asks to edit, cut, caption,...

  • 189 stars
  • 0 votes
  • 0 copies
  • 0 views
  • Added November 18, 2026
designgoaws

Works with

  • cli

Security analysis

A100/100

Pro scans all 3 files and shows the line behind each finding

Scanned November 18, 2026

npx -y skills add LemonFace0309/charles-skills --skill charles-claude-editing --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Charles Claude Editing?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Charles Claude Editing
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/lemonface0309-charles-claude-editing/badge)](https://www.skillsdirectory.com/skills/lemonface0309-charles-claude-editing)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: charles-claude-editing
description: Edit raw talking-head videos into engaging, organic-looking Instagram Reels / TikToks - cut dead space and stumbles, pick the best take, 2-3 word captions, retention-maxing image/video overlays (pop-culture stickers, cutouts, split screens, tweet screenshots), sound effects, a background track from the music folder, and a TikTok-title-style hook when asked. Covers news and education/entertainment videos. Use this whenever Charles shares a talking-head .MOV/.mp4 and asks to edit, cut, caption, add a hook, add overlays/b-roll/images/a split screen, add music or sound effects, "make it a reel", "minoify" it, or fix any of those in a video already edited this way - even if only one of those steps is asked for.
---

# Goal

* The goal of editing is to maximize viewer retention. The viewer should constantly have something new to look at, without ever feeling overwhelmed
* Keep it organic, like I made it myself on my phone. Use the real thing over a recreation of it: the actual post, footage, app or website, not a mock-up, stock image or AI version. If something looks like a template or AI slop, don't use it
* If something isn't covered here, do whatever best serves those two goals


# Decide these first

```
Is it news (layoffs, announcements, someone else's findings)?
├── Yes → also follow News Videos at the bottom. It wins wherever it disagrees with the defaults
└── No, it's education/entertainment → just the defaults

Did I ask you to open on a split screen?
├── Yes → open on one (see Choosing the Layout)
└── No → open on me
```


# General

* Cut any dead space that exists in the video. Stumbles, false starts and the moments after I finish a line all count (see Cuts)
* In the case that I filmed multiple takes of the same line, choose the best take
* Safe area (1080x1920): the IG/TikTok header covers the top 250 px, the username and caption sit below y 1500, and the buttons run down the right side below y 840. Keep text at least 80 px from the sides, 140 px below y 840. Be aware of these areas and don't put key information in them, but don't cut off or shift videos and pictures in a weird way because of them. Footage and images can still extend into those areas
* No gimmicks: no pop-ins, rounded corners (except for the hook title), black-background splits or emoji tiles
* My framing is usually close, with my face filling the middle of the frame. In that case never zoom or punch in on my face. Only add slow zooms between cuts when my face is smaller than the middle third of the frame. Cover jump cuts with B-roll instead
* Never isolate or denoise my voice. It struggles to do that well
* If I give you a reference reel, copy its structure, pacing and B-roll, but never its caption style. Captions always follow this file


# Cuts

* Cut on the picture as well as the sound. The moment I finish a line I start to relax (ex: looking down at my script, dropping a hand, reaching to stop the recording), and even 0.3 s of that makes a cut look sloppy
* End each clip within 2 frames at 30 fps (0.07 s) of my voice stopping, with the whole last word kept, trailing "s" or "t" included
* If I start to relax before then, end on the frame before it starts. If it starts before the word is over, cut the picture there and start the next clip's picture that many frames early, so the rest of the word plays over it (an L-cut, up to 4 frames). If it starts any earlier, or during my last word, use another take of the line or cover it with B-roll. Never clip a word to hide a movement
* Start each clip 1 frame before I speak. If I'm still settling into place there, use another take of the line or cover those frames with B-roll
* So a join keeps at most 0.1 s of silence, with no breaths, clicks or lip smacks. Measure it from the sound, not from where the cut sits: a breath or a pause before my first word counts, and it shows me getting ready (this repository's tools/reelkit/check_gaps.py measures every join). A beat I hold mid-line while still looking at the lens (ex: before a punchline) is delivery, not dead space, so keep it
* The reel's last clip can run up to 0.15 s after my last word, as long as I haven't started to relax
* Before you show me a cut, look at the frames on both sides of every join and the reel's last frames (this repository's tools/reelkit/check_joins.py draws them). Redo any join where I visibly start to relax


# Hook Title (the text caption at the top)

* No hook title unless I explicitly ask for one
* When I do: Helvetica Neue Bold near the top of the video, black text on a white background with rounded borders, inside the 80 px side margins. Make it look like TikTok's native text with a background: each line is wrapped individually, so the background bends where a shorter line meets a longer one
* Two lines max. Three is fine when the hook won't fit in two at full size, or when a CAPS word gets its own line
* Usually keep it up for 3-5 s (my first sentence). It can stay longer as long as there's no overlay or B-roll that should replace it


# Captions

* Helvetica Neue Bold, ~66 px, white with a thin black outline and a soft shadow. Sentence case, apart from the comment keyword (below)
* Around y 1380, inside the safe area. If an asset sits there, move the captions above or below it, never over it. While a split screen is up, they go on the seam
* 2-3 words each line
* No animations/transitions
* Go over the captions yourself to make sure the ones coming back from a transcription service are accurate to the context of what I'm saying. Sometimes transcription services suffer with names too, so fix those
* Censor any words that tiktok may block the video for, in the captions and the hook (ex: say s*****e instead of suicide)
* Caption the comment keyword (the word I ask viewers to comment) in caps, inside quotes, with the word and its quotes in yellow and the usual black outline: Comment “SCREENSHOTS”. It's the one word viewers have to type, so it shouldn't blend in with the words around it. In the hook, write it in caps and quotes in the hook's own black (in this repository, reelkit's `keywords` does the captions)


# Image/Video Overlays

* Use these by default unless I explicitly ask not to
* Make sure every asset is relevant to the video. Don't add assets for the sake of adding assets. Loosely related photos/videos are okay in education/entertainment
* Placement, for overlays that don't take over the whole screen with a blurred background:
   * Where my face sits decides it. If my face is closer to the top, put them below my face and move the captions lower with them. If it's closer to the bottom, put them above my eyes
   * Never cover my eyes or the captions
   * Long screenshots can run off the bottom as long as the highlighted lines stay above y 1500
* Don't blur anything more than it needs, and show assets whole. Don't crop one so part of it gets cut off (ex: the side of a phone screen)
* Never leave black bars around an asset. If it doesn't fill its space (ex: a web page with side margins, a screenshot on a black background), fill the edges with a blur of that same page or asset, or something else related. It's more aesthetic
* You can also place me in front of an image, with an AI cutout of my body in the foreground covering part of it. It adds depth. Only do it when nothing relevant gets covered
* Size images like Instagram stickers rather than full-width slides, so I stay on screen. Tweets, screen recordings and code are the exception, they need to be big enough to read
* If given assets, crop/highlight/adjust them to keep only the relevant parts. Feel free to chop them up to make them more engaging
* If you're finding assets yourself via the web:
   * No stock images. Prioritize images that feel like they were taken by a regular person on their iPhone. Obviously there are exceptions like logos, etc
   * Anchor on pop culture and trends that the viewer would recognize. We want to invoke a familiarity bias
   * Don't create AI generated photographs unless I explicitly ask, because they create an uncanny feeling. Animations and graphics are fine
   * Screenshots of tweets and other social media are great
   * Make sure it's accurate to what I'm talking about (ex: the current Mac mini, not an older model)
* Web pages: dark mode, starting at the visual part, not the intro text. Keep the content inside the 80 px side margins. Light mode pages w/ text running into the screen edge look ugly
* For big claims, show the source (official release, headline) with the words I'm saying highlighted
* Make your own motion graphics and recordings instead of using templates. When I explain a concept, build a visual for it (text, diagrams, graphs, beautiful animations). This is especially potent in educational videos
* When I explain a term, a fix, a technique, a number or a mechanism, ask charles-explainers-assets for the visual at the slot's size, with its text area, my word times and the change words. Place what it returns, and put clicks on its cues
* When you show a list or sequence as a row of pictures, don't use icons. Use organic-looking images you find online, cropped into squares/rectangles
* Graphs and charts are okay. Screenshots of real ones from other sources are great. If you make one yourself, it has to look professional and clean, and be accurate and detailed, w/ well timed animations. A sloppy one comes across as AI slop
* If I read out a list, put it up as an overlay with blur, each item revealed as I say it (plain caption-style text, not a screenshot). If I read it slowly (more than ~4 s between items), take the overlay down between items and bring it back for the next one, so it stays engaging
* Timing:
   * Don't go more than ~4 s without something new on screen (a new asset, a cut, a highlight moving to the next line)
   * No asset stays up longer than 3 s, except videos, tweets or code I'm reading, complex assets, or something I keep referencing. Those stay up as long as I'm talking about them
   * Anything back to back (images, tweets, clips) is a hard cut with no gap in between. Even a brief stutter isn't back to back


# Choosing the Layout

* Every asset goes on screen one of five ways. Pick by what the asset is and what the moment is going for:
   * Regular overlay, no blur: images I just want on screen, which is most of them. Sticker-sized, above or below my face (see Image/Video Overlays)
   * Overlay with blur: an image that takes up a lot of space and has to be seen whole, or a list I'm reading out (each item revealed as I say it). Blur me behind it and centre it
   * Split screen 50/50: another video that's taller (ex: footage from somewhere else, a phone screen recording). It gets half the frame (1080x960) and my face gets the other half, at the top or bottom
   * Split screen 40/60: another video that's landscape (ex: a desktop screen recording, a 16:9 clip). It gets the top 40% at full width (1080x768) and my face gets the bottom 60%, like @opusjake's splits
   * A new screen: B-roll or a new vertical video we generate that should be completely immersive. It takes the whole frame (1080x1920) and cuts away from me
* In a split, fill the video's panel with real footage, never a black panel with a card in it
* Only make the whole video a split screen when I ask
* Add a transition into and out of a split (ex: a push)
* When I ask you to open on a split (ex: me reading a Reddit post, a video demo showing something cool, tech news with a demo), put my first sentence on the seam as hook text, then the 2-3 word captions on the seam while the split is up
* Motion graphics from the charles-arcade-assets skill fit any of these. Pick the layout first, then ask charles-arcade-assets for the asset at that size, with the area its text has to stay inside. You place what it renders
* Teaching visuals from charles-explainers-assets fit the same way: pick the layout and slot first, then ask for the visual at that size


# Music

* Pick the folder in this repository's assets/music that fits the mood of the video (chill, emotional storytelling, funny, motivational, suspenseful, upbeat, wholesome)
* Exposés and serious news always get suspenseful (conspiracy/suspense-building)
* Move the music around to fit the video better. It doesn't need to begin from the start, so skip quiet intros
* Keep the music about 8-9 dB under my voice while I'm talking. Ducking it harder than that makes my voice sound isolated


# Sound Effects

* Only on moments that earn one: a camera shutter when a screenshot appears, a pointer click on a highlight or tap, a soft whoosh on a big transition. Bias towards premium and clean sounds like these. Get them from free online sources or the clean ones in my folder (ding, discord ping, apple pay, metallic riser)
* Meme sounds from my folder (assets/sound-effects in this repository, ex: vine boom, Cinema Sins dun) at most once per video, and only on a punchline
* No more than ~6 per minute. Too much and it gets annoying, but adding them in a timely and tasteful way improves the quality of the video
* Keep them at or under my voice level


# Export

* Small text anywhere on screen (a UI demo, a website, code, a tweet, a screenshot) means a 2160x3840 export. At 1080 wide, ~10 px UI text goes soft once the phone stretches it to fill the screen, and it looked blurry in Photos
* Capture screens at 2x and keep them at 2x the whole way. Never shrink a capture to 1080 wide and let it get scaled back up. Draw captions and the hook at 2x too. My footage is 1080x1920, so it gets scaled up once and that's fine
* Every position in this file is on the 1080x1920 grid. At 2160x3840, double it
* H.264, CRF 16 at 2160x3840 (about 25 Mbps), AAC 256k. Less smears fine text and grain
* A talking head with only captions and the hook can stay 1080x1920
* Save every finished reel to my Apple Photos. I post from my phone, so that's where I check it. Replace an old version rather than leaving both


# News Videos

* If I ask for a hook title, put the key words in CAPS, in larger text than usual (ex: "Samsung EXPOSED to be SPYING on users", or "Microsoft offers voluntary / TERMINATION / to 60,000 amid AI adoption")
* Mostly screenshots of the posts I send you, big (840 px wide) and held while I talk about them
* Yellow highlight on the words I'm saying, moving as I talk. It has to sit exactly on whole words
* No loosely related photos/videos

Files in this skill

  • README.md1.3 KB
  • SKILL.md14.2 KB
  • examples/reels.jpg96.4 KB

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

View all in design →

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.