In the Skill's Own Terms
Orchestrates product launch and promo video creation through a seven-step workflow. Takes a product URL, pasted script, or brief and produces an MP4 via HyperFrames.
Captures brand assets from live sites, adopts a frame preset and remixes it onto brand tokens, writes a storyboard with time-coded shot sequences paced to voiceover, generates narration and BGM, dispatches parallel sub-agents to build each frame as HTML, and renders after lint/check/snapshot verification. Targets SaaS promos, feature reveals, app launches, and site tours. Works signed in to HeyGen or offline with local TTS/BGM engines.
What it produces
- Final MP4 render at renders/video.mp4 with stated duration
- Contact sheet of snapshots at frame midpoints and cuts for verification
- STORYBOARD.md with time-coded shot sequences and SCRIPT.md when narration is needed
How It Works
- 01Confirm Brief and Initialize Project
The intent layer locks the brief before this workflow starts.
- 02Capture Assets from URL
Runs hyperframes capture on the product URL to collect screenshots, DOM structure, brand tokens (colors/fonts), visible text, and asset descriptions.
- 03Adopt and brand-remix a Frame Preset
Picks one shipped frame preset whose look fits the brand, then runs build-frame.mjs to copy the preset's FRAME.md, remap brand colors/fonts onto the preset's structure, and copy its caption skin.
- 04Write Storyboard with time-coded Shot Sequences
Drafts STORYBOARD.md and SCRIPT.md following story-spine doctrine.
- 05Generate Audio in Background
After storyboard approval, starts TTS narration with the chosen provider and voice, fetches BGM from HeyGen's music library by mood, generates word timings, and continues to visual design without waiting.
- 06Dispatch sub-agents to Build Frames in Parallel
Waits for audio to finish, syncs durations, then builds one bounded packet per frame and dispatches a sub-agent for each.