Skip to content

demo-video

Record and post-produce screen demos. Capture → trim → brand → narrate → music → final MP4.

  • Area: Foundational
  • Category: documents
  • Level: advanced
  • Input: conversation
  • Output: document
  • Setup: install
  • Roles:

Records and post-produces screen demos with ffmpeg and Remotion.

How you ask

  • “Record a demo of fix my marketing”
  • “Capture me running the prospect audit”
  • “Start recording a demo called daily-briefing”
  • “Record the board of advisors session”

Inputs

  • name — Demo identifier in kebab-case (e.g. fix-my-marketing)
  • name — Demo identifier in kebab-case
  • prompt — The prompt or skill to execute while recording
  • recording — Recording used to remove idle segments from a raw recording using frame-diff analysis
  • fade — Crossfade duration between segments (default: 0.5s)
  • recording — Raw or trimmed recording containing confidential content
  • spec — List of regions - x, y, w, h in source pixels, plus optional start/end seconds and strength
  • config — demo-videos.json with branding and per-demo messaging
  • demo — Demo ID from config (e.g. audit-before-after)
  • narration — Full narration text or @file.txt path
  • voice — TTS voice (Edge TTS default: en-US-JennyNeural; with --tts openai the default is echo)
  • proof-artifact — Screenshot to display as proof overlay (5s before outro)
  • source (Teaser (Hyperframes)) — Finished demo recording (e.g. {company.marketing.demo-recordings}/public/*.mp4)
  • scenes — 4 (start_seconds, duration_seconds) tuples for the 4 overlay-card scenes
  • brand — brncx | gbl
  • narration — Narration text following template in references/teaser-settings.md
  • name — Short identifier in kebab-case (e.g. thirdbrain-dashboard-tour)
  • script — Interaction script (.ps1, .py, or .js/.ts). Annotate with # CAPTION: "text" to auto-generate captions.
  • avatar_video — Talking-head MP4 from avatar-video skill (512×512 MuseTalk output or larger). Avatar audio is the sole audio track.
  • brand — a key of shorts-config.json _assets.branding: default (neutral, the default) | brncx | gbl
  • capture_mode — monitor (default, full-screen via record_demo.ps1) | playwright (browser session via playwright-cli)
  • captions — Pre-existing captions JSON to override auto-generation (for re-renders after tweaking)

Outputs

  • Per-brand palette, fonts, intro/outro copy and template key — created only when adding a new brand, then passed to the renderer as --brand.
  • Per-demo scene, narration and outro-headline config — created only when adding a new demo, then passed to the renderer as --demo on every later run.
  • blurred (mp4)
  • captions_json (json)
  • final (mp4)
  • prefer a permanent directory)
  • recording (mp4)
  • teaser (mp4)
  • timeline (json)
  • trimmed (mp4)
  • {output-dir}/final/{name}-narrated.mp4
  • {output-dir}/transcripts/{name}-timeline.json
  • {output-dir}/trimmed/{name}.mp4
  • {output-dir}/{name}-blurred.mp4
  • {projects-sibling-dir}/hyperframes-engine/renders/{demo-id}-{brand}.mp4
  • {scope.marketing.demo-teaser}/brand.yaml
  • {scope.marketing.demo-teaser}/{demo-id}.yaml