agentleFS
dxdleadydxdleady/playwright-demo-video/skills/demo-video/SKILL.md

demo-video

Skill0 starsChanged 24 days agoMITView on GitHub

Record polished product demo videos of your own web app with Playwright: scripted clicks, a smooth drawn cursor, DOM zoom that keeps text sharp, onboarding-style captions, spotlight, auto-trim, optional royalty-free music, exported as MP4 or GIF for docs, landing pages, READMEs and social. Demos are code, so when the UI changes you re-run one command instead of re-recording. Use when the user wants a product demo, a feature walkthrough, a screen recording, a GIF for a README, a clip for docs/landing/LinkedIn/X/TikTok, or says 'record a demo', 'make a GIF of X', 'show this feature in a video', 'снять ролик', 'сделай гифку'. Also use when a UI change means existing demos must be re-recorded. Not for editing user-supplied footage: this records the app itself.

27 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use karpathy-coder, ai-act-readiness and 10 more.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • csiddhant796-blipSkill

    karpathy-coder

    Use when writing, reviewing, or committing code to enforce Karpathy's 4 coding principles — surface assumptions before coding, keep it simple, make surgical changes, define verifiable goals. Triggers on "review my diff", "check complexity", "am I overcomplicating this", "karpathy check", "before I commit", or any code quality concern where the LLM might be overcoding.

    Also in 4 projects that use this

  • Hussain-ElfalahSkill

    ai-act-readiness

    Also in 3 projects that use this

  • Hussain-ElfalahSkill

    c-level-agents

    Also in 3 projects that use this

  • csiddhant796-blipSkill

    self-eval

    Honestly evaluate AI work quality using a two-axis scoring system. Use after completing a task, code review, or work session to get an unbiased assessment. Detects score inflation, forces devil's advocate reasoning, and persists scores across sessions.

    Also in 4 projects that use this

  • +8 more from Hussain-Elfalah
  • Other versions27 projects use a version of this.

    Word for word · 0

    None found.

    Edited · 0

    None found.

    Wrote their own version · 27

    1. liaohch3/claude-tap★ 3.3ksame name
    2. microsoft/hve-core★ 1.6ksame name
    3. alinaqi/maggy★ 711same name
    4. alinaqi/claude-bootstrap★ 708same name
    5. ooiyeefei/ccc★ 497same name
    6. getsentry/junior★ 409same name
    7. hmislk/hmis★ 244same name
    8. Kewton/CommandMate★ 44same name
    9. Hassaan146/claude-skills★ 18same name
    10. yzhao062/agent-config★ 16same name

    17 more not listed

    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.

The file

We split this skill into the rules it gives an agent

6 rules. 0 are common, 3 are distinctive, and 3 are very specific to this project.

  • 0

    Common

    Most of the 25 edited versions we read keep these.

  • 3

    Distinctive

    This author's own choices. Worth a look.

  • 3

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Demos are Playwright specs: When the UI changes you re-run a command instead of re-recording by hand, so the demos in your docs can't …

    line 13
  2. Very specific:

    Add to package.json scripts:

    line 33
  3. Distinctive:

    Delete it once the user's own demo works.

    line 49
  4. Distinctive:

    Check that the first frame is the app and not a loader, that the zoom actually happened, that every caption says something true, and that …

    line 111
  5. Very specific:

    Social reels (9:16, voiceover, music, titles): use the recordings as footage in HyperFrames.

    line 115
  6. Very specific:

    Placing in docs: write alt text as prose that carries the whole point.

    line 117

“Common” is measured: kept by at least 6 in 10 of the 25 edited versions of this file we could read. “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

Demo video recorder

Demos are Playwright specs. When the UI changes you re-run a command instead of re-recording by hand, so the demos in your docs can't quietly go stale.

What it adds on top of Playwright's built-in video, which shows clicks happening by themselves in an unzoomed full window:

  • a drawn cursor that glides and ripples on click (Playwright never moves the real pointer);
  • DOM zoom toward the element that matters, so text stays sharp unlike a crop in post;
  • coach-mark captions ("Step 2 of 3 · Delete it") and a spotlight that dims everything else;
  • auto-trim of the page loading before the demo and of the cleanup after it;
  • MP4 + GIF export with the ffmpeg flags people forget, plus an optional music bed mixed to sit behind the picture.

Install into the user's project (first time)

Needs @playwright/test and ffmpeg. From the project root, with SKILL = this skill's folder:

mkdir -p e2e && cp -r "$SKILL/templates/e2e/demo" e2e/demo
cp "$SKILL/templates/playwright.demo.config.ts" .
mkdir -p scripts && cp "$SKILL/scripts/"*.sh scripts/ && chmod +x scripts/*.sh
printf 'demo-out/\ne2e/demo/.auth/\n' >> .gitignore

Add to package.json scripts:

"demo": "playwright test --config playwright.demo.config.ts",
"demo:render": "./scripts/render-demo.sh"

Then set DEMO_BASE_URL (default http://localhost:3000) and optionally DEMO_DEV_COMMAND (e.g. npm run dev) and DEMO_MEDIA_DIR (e.g. public/docs-media). Details: references/setup.md.

Smoke test first. e2e/demo/todomvc.public.ts runs against Playwright's public TodoMVC, with no login and none of the user's code. Run it before writing anything for their app:

DEMO_BASE_URL=https://demo.playwright.dev npx playwright test --config playwright.demo.config.ts --project=public
./scripts/render-demo.sh --gif      # → demo-out/media/todo-basics.mp4 + .gif

Delete it once the user's own demo works.

The pipeline

npm run demo                              # record  → demo-out/raw/**/*.webm
npm run demo:render                       # encode  → $DEMO_MEDIA_DIR/<name>.mp4   (--gif for a GIF too)
./scripts/add-music.sh <name> track.mp3   # optional: your own track…
./scripts/add-music.sh <name> --mubert    # …or a generated royalty-free bed (Mubert API)

Iterate on one demo: npx playwright test --config playwright.demo.config.ts --grep "todo basics".

Writing a demo

One file per demo: e2e/demo/<name>.demo.ts (needs login, so it runs after auth.setup.ts) or <name>.public.ts (no login). The test title, kebab-cased, becomes the file name, and cue()/cut() must use that same name.

import { test } from '@playwright/test';
import { beat, glideAndClick, mountCursor, say, spotlight, zoomOut, zoomTo } from './camera';
import { clock, cue, cut } from './timeline';

test('invite a teammate', async ({ page }) => {
  const t0 = clock();
  await page.goto('/settings/team');
  await page.getByRole('button', { name: 'Invite' }).waitFor();

  await cue('invite-a-teammate', t0);          // everything before this is trimmed
  await mountCursor(page, { accent: '#0EA5E9' });

  await say(page, { step: 'Step 1 of 2', title: 'Invite by email' }, 1200);
  await glideAndClick(page, 'button:has-text("Invite")');
  // …
  await zoomTo(page, '.seat-count', 1.8);
  await say(page, { step: 'Step 2 of 2', title: 'Seats update instantly', note: 'Billing follows *automatically*.' }, 2000);
  await zoomOut(page);

  await cut('invite-a-teammate', t0);          // everything after this is trimmed
});

Full helper reference, the traps and how to restore data: references/writing-demos.md. Read it before writing the first spec for an app.

Rules that make it look produced

  1. Show the proof, not the click. Ask what a sceptic would need to see, and point the camera at it. For "delete the text and the audio goes with it", the proof is the duration readout dropping (33:25 → 33:14), not the deleted words. Find the number that moves.
  2. Hold after every action (beat, the hold argument of say). The instinct is to keep moving. Resist it.
  3. Zoom on intent, not on movement. Zoom when the point is one specific control, then hold. Zooming on every click is what makes cheap tools look cheap.
  4. One idea per clip. 6–12 s for a docs page, under 40 s for social.
  5. Real data, then put it back. Record against a real seeded account, because an empty app proves nothing. If the demo edits data, snapshot it first and assert it was restored afterwards (restore.ts).
  6. Captions match the product. Pass the product's accent colour and font to mountCursor. Don't invent a second visual language.
  7. MP4, not GIF, wherever <video> is allowed: about 7× smaller and sharp. Use GIF for GitHub READMEs, email and Slack.

Verify before you ship

Watch it; don't assume. Pull frames and look at them:

ffmpeg -y -i demo-out/media/<name>.mp4 -ss 8 -vframes 1 /tmp/frame.png   # then open the PNG
ffprobe -v error -show_entries format=duration -of csv=p=0 demo-out/media/<name>.mp4

Check that the first frame is the app and not a loader, that the zoom actually happened, that every caption says something true, and that the thing promised is visible on screen. A demo that shows the wrong state is worse than no demo.

Going further

  • Social reels (9:16, voiceover, music, titles): use the recordings as footage in HyperFrames. See references/reels.md.
  • Music: references/music.md.
  • Placing in docs: write alt text as prose that carries the whole point. It's what screen readers and AI agents read instead of the pixels.

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.