Product Demo Skill — FeedSkills

What you get

1080p portrait or landscape MP4 and a cover image
Captions, source notes and editable JSON/HTML/animation
A local project with stable scene IDs for revisions
layout.json with real SVG/DOM measurements for Feed output

You will need

Node.js 22+, FFmpeg and FFprobe
English copy and traceable inputs for each scene
First render downloads pinned HyperFrames 0.8.181 and a browser
Local Kokoro for default narration; explicit silent export is supported

Good fit

A real webpage or software feature with before, action and result captures
Readable walkthroughs with verifiable product behavior

Not for

Automatic capture of private pages without user access
Simulated interfaces presented as working software

You say

“Use the bundled workflow to make a video from my topic and sources. Check missing inputs, preserve evidence and show a preview before export.”

After installing, say something like this to your agent with your own topic and assets. It asks only for what is missing.

FAQ

Do I need an account?

No. Copy the install prompt into your agent, or download the free ZIP.

What does it cost to run?

The Skill is free. It needs the tools listed under “You will need”. Any voice, model or media service you choose bills you directly; FeedSkills does not include provider credits.

Can I use the videos commercially?

Videos you make are yours to use, including commercially. The Skill package is licensed as: MIT. Fonts, music, footage and services you add keep their own terms.

Which agents does it work with?

Agents that read SKILL.md files, such as Codex or Claude Code. What FeedSkills has actually tested is listed under Tested scope below.

Tested scope

Version: 0.1.1
Acceptance: This version has a recorded acceptance entry.
Tested: Two 16-second Feed demos were human-approved. Real DOM measurement, explicit local Kokoro AAC mixing, focus crops, silent-export and atomic-output regressions passed. Fresh-OS and physical-device checks are not claimed.
Runs on: Codex · macOS · Node.js 22+ · FFmpeg
License: MIT

Details

Show a real state change

Product Demo keeps your real software captures visible while adding a readable headline, focused detail, local English narration and burned captions. A screenshot sequence shows recorded states; it does not prove continuous cursor timing or a causal click.

Watch both 16-second examples · Download the free Skill. Both current cuts passed human review. Fresh-system and physical-device acceptance remain separate.

Try a complete editable project

The example ZIPs include captures, narration and the local renderer. Extract one, enter its top-level folder, then run npm run check, npm run preview and npm run render. Each fresh example omits an existing MP4 so the renderer can create a new output.

The current examples use traceable local UI captures from October 2026, not a continuous recording or a claim about current production behavior. Inspect their input and source records before adapting them. The brief-only files in the Skill need your own captures.

Make your own

Install with your agent, supply a topic and rights-cleared screenshots or a recording, then create a new project outside the installed Skill folder. Feed scenes last 1.5–4 seconds and the film totals 15–60 seconds. Legacy 16:9 inputs remain supported.

Local Kokoro narration is the default. voice:false, silent:true or --no-voice makes a silent export; narration:false silences one scene. A failed requested narration stops the render. Review the exported picture and audio before delivery.

The package is MIT-licensed. Rendering uses Node.js 22+, FFmpeg and pinned HyperFrames 0.8.181; it requires no paid media generation service. Agent-provider usage and access to private capture pages remain your responsibility.