changelog-video

Converts weekly changelog markdown into branded square videos with voiceover, captions, and animated UI mockups.

43.5k|4.2k|Updated Mar 10, 2026
One-click install
npx skills add https://github.com/heygen-com/hyperframes --skill changelog-video
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: changelog-video
Source: https://github.com/heygen-com/hyperframes/tree/main/.agents/skills/changelog-video
Command: npx skills add https://github.com/heygen-com/hyperframes --skill changelog-video

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) and assets (resource) components.

What problem does it solve?

Turning a weekly product changelog into a polished, on-brand video normally requires manual scripting, voiceover recording, motion design, and caption timing. This Skill automates that entire pipeline, producing a lint-clean HyperFrames project from a changelog markdown file.

Core Features & Use Cases

  • Changelog-to-Video Pipeline: Parses a changelog .md, makes editorial cuts to fit a 45-60s budget, and routes each theme to an animated UI mock, terminal scene, or checklist via a visualization registry.
  • Two-Layer Script and Voiceover: Writes token-based scripts separating display spelling from spoken phonetics, generates HeyGen TTS voiceover with word timestamps, and aligns captions automatically with a fallback to Whisper forced alignment.
  • Branded Scaffold and Quality Gates: Ships bundled fonts, background video, background music, and a master HTML skeleton enforcing brand tokens, plus hard gates for lint checks, seam verification, and caption presence.
  • Use Case: Given a weekly digest markdown with 30 changelog items, produce a 1080x1080 square video with Annie voiceover, word-timed captions, animated mock-UI scenes, and a digest-link outro, ready to render on request.

Quick Start

Use the changelog-video skill to turn my weekly changelog markdown file into a branded 45-60 second square video with voiceover and captions.

Frequently Asked Questions about changelog-video

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I turn a changelog markdown file into a video?▼

Provide the changelog .md to the skill, which parses themes and items, cuts them to a 45-60 second budget, routes each theme to an animated UI mock or terminal scene, generates voiceover and captions, and builds a HyperFrames project ready to render.

How are video captions generated and timed to the voiceover?▼

The skill writes a two-layer script separating display spelling from spoken phonetics, generates HeyGen TTS audio with word timestamps, then runs align-captions.mjs to map spoken timings back onto display tokens. If timestamps are missing, it falls back to Whisper forced alignment.

What voice and format does the changelog video use?▼

The video is square 1080x1080, roughly 45-60 seconds, using the pinned HeyGen Annie voice, bundled TT Norms and ABC Solar fonts, an animated background pattern, and a 32px caption rail at the bottom.

What happens if the TTS word timestamps are missing or empty?▼

Missing word timestamps are a hard gate; the build cannot proceed without them. The fallback runs local Whisper with word_timestamps enabled on the produced audio, then feeds that JSON into the caption aligner using display spellings.

Does the skill render the final MP4 automatically?▼

No. The skill builds and verifies a lint-clean HyperFrames project but only renders when the user explicitly asks. After a requested render, it verifies frames from the MP4 for captions, background video, and frozen frames.