agentleFS
glifxyzglifxyz/glif-mcp-server/skills/app-demo-videos/SKILL.md

app-demo-videos

Skill213 starsChanged 40 days agoMITView on GitHub

Record a screencast of an app you are building and turn it into a polished demo, explainer or promo video with Glif, using video templates for device mockups, captions, voiceover and music. Use when: "make a demo video", "record a screencast", "I need a launch video for my startup", "hype trailer for my app", "product walkthrough", "feature tour", "explainer video", "turn my Loom into a polished demo", "show my app on a MacBook", "scroll through my dashboard and zoom in", "15-second TikTok ad for our best feature", "changelog video", "App Store preview video", "GIF for the README", "thumbnail for the launch video". NOT for: still graphics (use app-promo-materials), videos unrelated to the app (use audio-and-video), animating a single image (use audio-and-video).

  • Installs packages

The file

We split this skill into the rules it gives an agent

7 rules. About 0 are common, 2 are distinctive, and 5 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈2

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 5

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Use the user's own recording or Loom export if they have one.

    line 32
  2. Distinctive:

    Start the app locally and seed it with demo data.

    line 34
  3. Very specific:

    Run it with DEMO_URL=<local url> node record-demo.mjs.

    line 37
  4. Very specific:

    Check the result, or pull frames with ffmpeg -ss <seconds> -i <file> -frames:v 1 frame.png.

    line 38
  5. Very specific:

    If ffmpeg is installed, convert to MP4: ffmpeg -i in.webm -c:v libx264 -pix_fmt yuv420p -crf 18 demo-raw.mp4.

    line 39
  6. Very specific:

    Keep recordings out of git, such as in recordings/ listed in .gitignore.

    line 41
  7. Very specific:

    Call compose_project once with:

    line 47

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 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

Record the real app, then let Glif frame it in a video template. The user's explicit instructions come first, within host safety rules.

1. Plan the video

Agree on a short plan before recording:

  • Story: 3 to 6 beats, one feature or action each. Open with the most useful moment; social cuts need a hook in the first 3 seconds.
  • Length: 15 to 30 seconds for social and teasers, 30 to 90 for walkthroughs and explainers.
  • Template: pick one, or describe your own.
    • Browser tour: the app in a clean browser window on a soft background, zooms on clicks, a caption per beat.
    • Device composite: the real recording placed on a laptop or phone screen in a generated lifestyle scene. Glif tracks the screen if the camera moves.
    • Social ad demo: vertical, 15 seconds, hook first, big captions, music.
    • Narrated walkthrough: voiceover and captions over the browser tour.
    • Explainer: animated slides in a chosen style, narrated, with short clips of the real app. Good for "how it works".
    • Changelog video: release notes or a changelog URL turned into a short narrated update.
    • App Store preview: real app footage only, 15 to 30 seconds, vertical. Check App Store Connect for current specs.
  • Outputs: 16:9 master at 1920×1080, plus a 9:16 1080×1920 cut if wanted, a poster image, a 1280×720 YouTube thumbnail or a README GIF.
  • Sound: voiceover script, voice (Glif can play samples to choose from, or clone a voice from a sample with the owner's consent), music mood (ducked under the voice), and optional click sounds synced to the cursor.
  • Captions: word-by-word highlight, simple lower thirds, or motion-graphic overlays with numbers and charts. Ask Glif for a still frame with a few font options first if the look matters.

Show the plan and wait for a yes. Video uses more credits than images.

2. Record the app

Use the user's own recording or Loom export if they have one. Otherwise:

  1. Start the app locally and seed it with demo data. Never record real customer data, emails, tokens or keys.
  2. Copy scripts/record-demo.mjs into the project's scratch space. Install Playwright if the project lacks it (npm i -D playwright and npx playwright install chromium), after checking with the user.
  3. Replace the example beats with the planned steps. glideClick moves a visible cursor before clicking. smoothScroll scrolls gently. Leave about a second of stillness after each beat.
  4. Run it with DEMO_URL=<local url> node record-demo.mjs. Add DEMO_MOBILE=1 for a phone layout. It prints each beat's time; keep those for captions.
  5. Check the result, or pull frames with ffmpeg -ss <seconds> -i <file> -frames:v 1 frame.png. Re-record if something glitched.
  6. If ffmpeg is installed, convert to MP4: ffmpeg -i in.webm -c:v libx264 -pix_fmt yuv420p -crf 18 demo-raw.mp4.

Keep recordings out of git, such as in recordings/ listed in .gitignore.

3. Send it to Glif

Upload each recording and the logo with upload_file: pass filename and kind: "video" (or "image"), then run the returned command with the real path. Attach the returned fileUrl values.

Call compose_project once with:

  • the chosen template, length and every output size
  • the beat list with timestamps and exact caption text, in quotes
  • the voiceover script, voice choice and music mood
  • brand colors (hex), logo and an intro or outro card (app name, one line, URL)
  • this rule: "Keep the app footage real. Frame, crop, zoom, speed up and caption it, but never regenerate or restyle the interface. Use a programmatic video template, such as Remotion, so captions and timing are exact."

Share the projectUrl. Follow the run with get_job_status, passing the jobId as job_id. A video can take several minutes; keep working on other things meanwhile. A pending status or timeout is not a reason to start another run.

4. Check and save

Download with curl -fsSL "<media url>" -o <path> into marketing/video/, or public/videos/ if the site embeds it. Check:

  • size, length and codec: ffprobe -v error -show_entries stream=codec_name,width,height:format=duration -of csv=p=0 <file>
  • caption spelling and timing against the beats
  • the interface looks like the real app

Fix problems in the same project by passing its project_id. Don't start a fresh run for a small fix. Useful follow-ups there: a 9:16 cut, 3 alternative 3-second openings to test, or a thumbnail.

To embed on the site:

<video src="/videos/demo.mp4" poster="/videos/demo-poster.jpg" autoplay muted loop playsinline width="1920" height="1080"></video>

For a README, a short MP4 works on GitHub. A GIF should stay under about 10 MB; ask Glif for one or make it locally with ffmpeg.

List the files you made. Posting or uploading to stores is the user's call.

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.