agentleFS
datnpqdatnpq/gskill/skills/ad-assets/SKILL.md

ad-assets

Skill0 starsChanged 48 days agoMITView on GitHub

Produce a full Performance Max creative set from a live website without a designer — mobile screenshots via headless Chrome, branded scene images in 1.91:1, 1:1, 4:5 and 9:16, logo sizes, and three short videos (16:9, 1:1, 9:16) rendered with ffmpeg (Ken Burns zoom + crossfades). Use when a campaign needs images or videos, "set đủ video", or ad strength is below "Excellent".

The file

We split this skill into the rules it gives an agent

1 rule. About 0 are common, 0 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈0

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Check a frame from each before uploading: ffmpeg -ss 7 -i brand-ngang.mp4 -frames:v 1 check.png.

    line 55

≈ “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

Ad assets from a live site

Real screenshots of the product outperform stock art and pass review easily. Pipeline:

capture_screens.mjs  →  shots/*.png          (mobile screenshots of real pages)
compose_scenes.py    →  scenes/<format>-N.png (branded frames, 3 formats)
render_video.py      →  <name>-<format>.mp4   (8 scenes × ~1.85 s ≈ 15 s)

Requirements: Chrome/Chromium, Node ≥ 22 (built-in WebSocket), Python 3 + Pillow, ffmpeg.

1. Screenshots

Edit a JSON config (see ${CLAUDE_PLUGIN_ROOT}/templates/assets.example.json) with the pages to capture (URL, optional scroll offset) and any localStorage keys that dismiss onboarding tours, then:

node ${CLAUDE_PLUGIN_ROOT}/scripts/capture_screens.mjs assets.json out/shots

Uses 390×844 @2x iPhone emulation. Use demo data in URLs, never a real customer's.

2. Scenes

python ${CLAUDE_PLUGIN_ROOT}/scripts/compose_scenes.py assets.json out

Each scene = eyebrow line, title, subtitle, and either a screenshot in a phone frame or the logo (first and last scene). Last scene carries the call to action and the domain. Colors and fonts come from the config — use the site's own palette so ads look like the site. Fonts must contain the language's diacritics (for Vietnamese: Segoe UI, Be Vietnam Pro, Noto Sans all work).

Still images for the asset group: export scenes 0 (brand) and 2–3 feature scenes in each ratio; 1.91:1 → 1200×628, 1:1 → 1200×1200, 4:5 → 960×1200 (resize the portrait scene).

3. Videos

python ${CLAUDE_PLUGIN_ROOT}/scripts/render_video.py out/scenes out/video --name brand

Per format: every scene gets a slow 5 % zoom (zoompan), scenes join with 0.35 s crossfades (xfade), H.264 yuv420p, 30 fps. Output ≈ 15 s, 3–5 MB each — above Google's 10 s minimum and small enough to upload quickly.

Check a frame from each before uploading: ffmpeg -ss 7 -i brand-ngang.mp4 -frames:v 1 check.png.

4. Logo

Square 1200×1200 (icon on a padded background, not cropped) and landscape 1200×300 (icon + wordmark). PNG. The brand name in the logo must match the business name field.

Upload notes

  • The Chrome extension file upload tool takes ≤ 10 MB per call — keep videos small.
  • Videos go to YouTube behind the scenes; pick the Video ad upload channel, not a personal channel (see pmax-campaign).

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.