agentleFS
binay-pybinay-py/claude-skills/skills/gif-maker/SKILL.md

gif-maker

Skill1 starsChanged 6 days agoMITView on GitHub

Make sharp, small, attractive animated GIFs (plus MP4/WebP) for READMEs, docs, social posts, Slack and product pages — from an HTML/CSS/JS animation (recorded frame-by-frame on a virtual clock, so motion is perfectly smooth and identical every run), a scripted terminal session (typed commands + real output), a video clip (trim, crop, speed, caption) or a set of images (slideshow with transitions). Two-pass palette, good dithering, and an automatic size budget (default 5 MB) that lowers fps → colors → width until it fits. Use whenever the user wants a GIF, an animated demo, a README animation, a terminal recording, or to shrink a GIF.

  • Deletes or force-pushes

3 other projects use a version of this.

Adopted by

Most recently taken up by

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
  1. Distinctive:

    Use real output: Run the commands first and paste what they printed (trim paths, never include secrets).

    line 30
  2. Very specific:

    Motion design: Use easing (cubic-bezier), stagger elements by 60–120 ms, and keep movement to 1–2 things at a time.

    line 41
  3. Very specific:

    Check it: Look at a few frames (ffmpeg -i out.gif -vf "select='eq(n\,0)+eq(n\,30)',tile=2x1" -frames:v 1 check.png) before calling it done.

    line 46

≈ “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.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

gif-maker

scripts/gif.mjs is zero-dependency Node 22+ and needs ffmpeg. html and terminal modes also launch a headless Chrome/Edge/Brave/Chromium with a throwaway profile (set BROWSER= to choose).

node gif.mjs html anim.html --dur 4 --size 800x450 --fps 24 --out hero.gif
node gif.mjs terminal session.txt --out demo.gif                  # --theme light, --size 900x520
node gif.mjs video screen.mp4 --start 3 --dur 5 --width 720 --speed 1.5 --text "One click" --out clip.gif --mp4
node gif.mjs images a.png b.png c.png --hold 1.6 --fade 0.4 --size 1000x600 --out tour.gif
node gif.mjs optimize big.gif --max-mb 3 --out small.gif

Common options: --max-mb 5, --fps, --colors 256, --width, --loop 0 (forever) / -1 (play once), --pingpong (html/terminal: play forward then back for a seamless loop), --mp4 / --webp (also write those), --keep (keep the PNG frames).

How each mode works

  • html: the page's clock is replaced before it loads. requestAnimationFrame, performance.now, Date.now, setTimeout/setInterval and all CSS/Web Animations advance exactly 1/fps per frame, and each frame is a screenshot. No dropped frames or jitter, and the same input gives a byte-identical GIF. Fonts and images are awaited first. Write the animation as a normal page; anything CSS/JS can draw works (canvas and SVG included).
  • terminal: a text script becomes a styled terminal window, rendered through the html mode.
    @title my-tool · demo
    $ npm run check            ← typed out character by character
    ✓ 12 files ok              ← printed (✓/ALLOW green, ✘/BLOCK red, !/WARN yellow, # dim)
    [yellow]custom color[/]    ← explicit colors: green red yellow blue dim bold
    @pause 900                 ← hold
    @clear  @speed 30  @prompt ❯
    Use real output. Run the commands first and paste what they printed (trim paths, never include secrets).
  • video: ffmpeg trim (--start/--dur), --crop W:H:X:Y, --speed, --text caption (bold, outlined, bottom-centered, shrunk to fit the width; Devanagari works, but start a mixed caption with the Nepali word because ffmpeg shapes a line in its first letter's script; emoji are dropped because ffmpeg can't draw them), then the same palette pipeline.
  • images: each image is fitted onto a --size canvas (--bg color) with --transition (fade, slideleft, wipeleft, circleopen… any ffmpeg xfade).

Making it look good (do this, not just "a GIF")

  1. One idea, 3–8 seconds. A GIF is a hook, not a tutorial. Show the before → after or the single magic moment, then hold the end state ~1.5 s so people can read it.
  2. Size for where it's shown. GitHub README: 800–960 px wide. Twitter/X and LinkedIn: 1200×675. Slack: ≤ 600 px and ≤ 2 MB. Mobile-first: square 1080.
  3. Readable text. At least 28 px headings and 16 px body at the final width; high contrast; dark terminal themes compress best.
  4. Frame rate by content. UI and terminals: 12–15 fps. Smooth motion and easing: 24–30 fps.
  5. Colors. Flat colors and simple gradients beat photos and noise, which bloat the file. Big smooth gradients can band in 256 colors; use --webp/--mp4 alongside for those, or add subtle texture.
  6. Loop cleanly. End where you started, use --pingpong, or design CSS animations whose duration divides --dur.
  7. Motion design. Use easing (cubic-bezier), stagger elements by 60–120 ms, and keep movement to 1–2 things at a time.
  8. Don't make it look AI-generated. Viewers have learned the tells, and they make a project look low-effort:
    • Avoid: purple-to-blue gradients and glowing "aurora" blobs; gradient-filled headline text (the fastest giveaway); emoji used as icons or section markers; rows of pill badges; frosted glass; three identical feature cards; everything centered; vague slogans ("supercharge your workflow").
    • Do: show the real product doing a real thing with real output (a terminal, the actual UI, a before → after); flat solid backgrounds; one accent color used for meaning (red = blocked, green = ok); a plain system or monospace font; left-aligned content like a real screen; specific text a competitor couldn't reuse ("blocked rm -rf ~ /old-build").
    • A plain screen capture of something real beats a polished motion-graphics banner. Use terminal --theme light or record the actual app with html <url>.
  9. Check it. Look at a few frames (ffmpeg -i out.gif -vf "select='eq(n\,0)+eq(n\,30)',tile=2x1" -frames:v 1 check.png) before calling it done.

Limits

  • GIF is 256 colors per frame; photos and video look worse than MP4. Offer --mp4 (GitHub READMEs can't autoplay MP4 from the repo, but issues, PRs and most sites can).
  • html mode captures the page as rendered by headless Chromium; WebGL works but can be slow per frame. Audio is ignored.
  • A terminal session's length is capped at 60 s (html and video take whatever --dur you give). GIFs that long should be videos.

Download now (ZIP)

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
What did you use it for? e.g. Reviewing pull requests in a Rails app

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.