ai-product-video

An agent skill for crafting cinematic product videos: 152 shot recipe cards · 209 styles · tuned Remotion implementations · a validated shot structure

English | 中文 | 日本語

ai-product-video is an AI agent skill that turns Claude Code or Codex into a
motion-design studio: point it at your product and it storyboards, animates, and
sound-designs a cinematic promo, marketing, launch, or demo video with
Remotion — real page captures, 2.5D camera moves,
beat-synced cuts, and film-grade SFX included.

🖼️ Browse the library — read the cards in references/shots/,
grouped into 10 functional categories.

✨ What's new

[!IMPORTANT]

🆕 2026-08 · 48 new shot recipe cards

The library grows from 104 to 152 cards / 209 style variants. Distilled from
209 candidate motions through eight rounds of frame-by-frame review against
reference footage, then folded into the regular categories with full
recipe cards, native Remotion components
(demos/<category>/<name>/<Component>.tsx, deterministic and driven by the
normalized progress t — see demos/README.md for the wiring snippet).
All de-branded: neutral placeholder copy and a single
swappable ACCENT color variable.

• 2026-07 · Audio library restructure — sfx/<category>/ with 149 SFX across
  16 scene/material categories, md5-deduplicated with license URLs restored.
  Music is not bundled here — it comes from dlazy search_audio.

🎬 Showcase

The 38-second library intro below was itself produced with this skill —
storyboard, shot implementation, and sound design were all done by an agent
following the toolkit's methodology:

https://github.com/user-attachments/assets/cba2df8a-4b2e-4247-bace-d0b1dea9c2bd

▶️ Watch in HD on YouTube

🚀 Quick start

The most direct way: hand the repo link to your agent.
In Claude Code / Codex or a similar agent, just say:

Install this skill for me: https://github.com/dlazy-ai/ai-product-video

The agent will clone the repo and link it into your skills directory. Or install
with the skills CLI / manually:

npx skills add dlazy-ai/ai-product-video

git clone https://github.com/dlazy-ai/ai-product-video.git
cd ai-product-video
ln -s "$(pwd)" ~/.claude/skills/ai-product-video   # Claude Code
# or
ln -s "$(pwd)" ~/.codex/skills/ai-product-video    # Codex

Set up the dlazy CLI

Visuals are generated by Remotion, but music, extra SFX, voiceover, captions
and B-roll come from the dlazy CLI — one command line
across many models, no per-vendor API wiring:

npm install -g @dlazy/[email protected]
dlazy login   # device-code flow, works over SSH
dlazy -h      # list every tool available to your account

dlazy login writes the key to ~/.dlazy/config.json. Prefer not to install
globally? npx @dlazy/[email protected] <command> works too. Already have a key?
dlazy auth set <key> — get one at
dlazy.com/dashboard.

The 149 bundled SFX need no account. You only need the CLI once the film wants
music, a voiceover, captions, or a B-roll shot that screenshots can't give.

Then make requests like:

Use ai-product-video to create a promo for my desktop product.
Use the deck-deal-flyin and row-embed shot cards to present this feature.
Design a product close-up inspired by spotlight-hero-card.

If no shot card is specified, the skill introduces the Ink Press shot structure
first and asks whether to use it; you can also pick shots from
references/shots/ before starting.

📼 Video template: Ink Press

The skill ships with Ink Press — a validated, complete promo template:
36.2 seconds, 1920×1080, 30fps, 10 shots in a paper-ink-amber style, with 2.5D
real-page camera moves, title cards, transitions, and a fully pinned cinematic
SFX pass:

https://github.com/user-attachments/assets/4cf5af51-98f3-4af2-8ab2-7267f470513d

▶️ Watch in HD on YouTube

To use it, just tell your agent:

Use ai-product-video to make a promo for my product with the Ink Press template.

The agent swaps in your product's screenshots, copy, and branding to reproduce
the same quality — the fastest, most reliable path to a finished film.

More templates are on the way.

Headless / CI notes

Rendering on a headless Linux box (tested: 2 cores, Node 22) hits three walls
worth knowing:

1. Concurrency cap — remotion still/render fails with "Maximum for
   --concurrency is 2" on low-core machines. Fix: pass --concurrency=1.
2. Old Headless removal — recent Chrome/Chromium dropped old headless mode;
   pointing Remotion at system chromium fails to launch. Fix: use a
   chrome-headless-shell binary instead of full Chrome.
3. Blocked CDN — if remotion.media is unreachable, the automatic
   headless-shell download is rejected. Fix:
   --browser-executable=<path-to-local chrome-headless-shell>.

With these three flags, frame renders work once your own captured assets are in place.

📦 What's included

The toolkit primarily targets web and desktop product promos, while individual
shot cards can also be used in feature demos, brand films, launch videos, and
other motion projects.

🗂 Repository structure

ai-product-video/
├── SKILL.md                 # Agent entry point and core production rules
├── references/
│   ├── pipeline.md          # End-to-end production workflow
│   ├── shots/               # 152 shot recipe cards in 10 functional categories
│   ├── sequences/           # Reusable full-video structures and sequence patterns
│   ├── aesthetic-rules.md   # Visual QA criteria
│   ├── music-beat-sync.md   # BGM analysis and beat-sync methodology
│   └── sound-design.md      # Sound-design guidance and examples
├── demos/                   # Remotion reference implementations (same categories)
├── template/                # Reference film: shot structure + scene source (assets not bundled)
└── assets/
    ├── lib/                 # Reusable Remotion components
    ├── scripts/             # Page-asset capture scripts
    └── audio/               # SFX only — BGM comes from `dlazy search_audio`
        └── sfx/<category>/  # 149 SFX across 16 scene categories

For the complete workflow and implementation requirements, see SKILL.md,
the production pipeline, and the
visual QA criteria.

🔊 Audio and asset notes

Audio files under assets/audio/ may be used according to their respective license terms.
See ATTRIBUTION.md for sources and license details.

SFX are organized into 16 scene/material categories (transition impact riser
camera ui text paper film light data scifi mech glass fluid
crowd counter) — pick the category first, then the timbre. See
sound-design.md for the category index and per-file usage.

assets/audio/ ships SFX only — BGM is fetched on demand with
dlazy search_audio, so no music files are bundled. Record the title, source, and
license of anything you pull in.

The reference film's screenshots are not bundled either: template/ ships its
shot structure and scene source, and you capture your own product's assets per
TEMPLATE.md. Before publishing, verify whether any product,
customer, or personal data in your captures needs to be anonymized.

🙏 Acknowledgements

ai-product-video is built on video-shotcraft by Wei Yihao
(Apache-2.0) — the shot card library, demo implementations, production pipeline,
and the Ink Press reference film all originate there. This version re-points
music, voiceover, captions and B-roll at the dlazy CLI and
drops material that isn't ours to redistribute; see NOTICE.md for
the complete list of changes.

Many shot recipes in this library were distilled by studying the motion language
of outstanding official product films — including promos from ClickUp,
Perplexity, Slack, Notion, Figma, Framer, Bear, Raycast, Pitch, Miro, Superhuman,
and Loom. The cards document motion techniques (timing, easing, choreography)
re-implemented from scratch; no footage, artwork, or brand assets from these
films are included in this repository. All trademarks belong to their respective
owners, and none of these companies are affiliated with or endorse this project.
Per-batch sourcing notes for the 48 cards added in 2026-08 live in
references/shots/ATTRIBUTION.md.

Special thanks to:

• Remotion — the React-based video framework
  that powers every demo and template here. Note that Remotion has its own
  license
  (free for individuals and small teams; companies may need a paid license).
• Mixkit — source of the 149 bundled SFX, under their
  free commercial license.
• The game-feel and animation communities whose published principles (e.g.
  Vlambeer's screenshake talks, classic animation timing) inform several cards.
• Claude Code — this library itself was built, iterated, and QA'd with an
  AI coding agent, using the same workflow the skill teaches.