How to use the Video ShotCraft agent skill (video-shotcraft)

What the video-shotcraft skill makes the agent do

Once installed, the agent can plan and build cinematic product promo videos. It works from real screenshots of a web or desktop app, uses a library of 124 reusable camera-motion shot cards, and can start from a finished template called Ink Press.

The skill also covers sound. It can find, add, replace and time sound effects for an existing product video or shot, including searching Freesound for candidate clips. It follows a multi-stage workflow with quality checks and asks you to choose a production mode before it produces anything.

• Full product promo videos from a frontend project or webpage
• Reproducing the Ink Press template with your own product content
• Single shot-card effects for any video
• Sound effect candidate searches, additions, replacements and alignment

Install the video-shotcraft skill

Download the skill ZIP from the video-shotcraft page on this directory, then unzip it. You can install it for yourself or for one project.

For all your projects, unzip it into ~/.claude/skills/. For a single project, unzip it into the .claude/skills/ folder inside that project. The folder should contain SKILL.md at its top level.

In the Claude apps, you can instead upload the ZIP under Settings → Capabilities → Skills.

• Personal install: ~/.claude/skills/
• Project install: .claude/skills/ inside your project
• Apps: Settings → Capabilities → Skills, then upload the ZIP

Check what you need before you start

The skill runs scripts and uses the network. The workflow relies on Remotion and ffmpeg, so it will not run as a plain text task. Plan for the following tools before you ask for a video.

Sound effect search needs Freesound API access. Music-synced videos need Python with librosa for beat analysis. Page captures need a headless browser.

• Node.js and npm
• Remotion (run through npx remotion)
• ffmpeg
• A headless browser for page captures
• Python with librosa, for beat analysis in music-synced videos
• Freesound API access, for sound effect search and download

Trigger the skill with natural requests

You do not need special syntax. Name the skill or describe the job in plain words. The skill is written to recognise requests to make a video, reproduce a template, make a single shot, or find, add, replace or align sound effects.

Before a full promo is produced, the agent checks your project or page with a short read-only review. It then offers three modes: use the Ink Press template, free creation where the agent decides the direction, or guided creation where you confirm key decisions. If you already name a mode, such as Ink Press, the agent uses it without asking again.

• "Use video-shotcraft to make a product promo for my web app."
• "Reproduce the Ink Press template for my product."
• "Make a single shot in the style of spotlight-hero-card."
• "Find sound effect candidates for the card flip in my product video. Do not download anything yet."

Choose a production mode

Template mode keeps the Ink Press structure and swaps in your screenshots, copy and brand details. Free creation lets the agent make the creative and engineering decisions and keep going to the final video. Guided creation pauses for your approval at the product brief, visual direction, shot mapping and final storyboard.

If you are unsure, pick guided creation for a first project so you can review the direction early. Choose free creation when you trust the agent to decide and want fewer interruptions.

• Template: follow the Ink Press replacement flow
• Free creation: the agent proceeds without stage-by-stage confirmation
• Guided creation: you confirm each key stage

Safety notes: scripts, network and data

The skill runs scripts and uses the network, including Freesound search and download. Review what it runs before you grant it access to a project you care about.

The workflow asks for real pages and captures, so it can expose anything shown on screen. If the page contains customer, personal, internal, secret or live data, the skill asks you to use fictional or redacted content and to freeze it before capture. Do not use this skill on confidential projects unless you have checked the screens first.

• Scripts are run on your machine and network access is used
• Use fictional or redacted data for any sensitive page
• Review the product check and storyboard before rendering

Tips for better results

Give the agent a clear product, a short list of key features and the audience. Name the shot cards you want if you already know them. The gallery at https://vincentwei1021.github.io/video-shotcraft/ shows live previews you can use to pick effects.

For music-synced videos, choose the music before the agent starts, so beat analysis can match transitions to the beats. For sound effects on an existing video, you can ask for candidates only and decide later whether to download.

• Name shot cards exactly as they appear in the gallery
• Pick the music first for beat-synced cuts
• Ask for candidate-only sound effect searches to review before downloading