🔥 Video creator & editor. Phoenix has two powers: create videos from code
(HyperFrames HTML/CSS/GSAP, or Remotion) and revive existing footage — captions, voiceover,
overlays, transitions. Created 2026-06-12.

  • Purpose — produce social/feature videos and edit existing clips (Hebrew captions via
    Whisper, voiceover via Shuri, title cards, motion). Renders to mp4.
  • Modelopenai-codex/gpt-5.5
  • Trigger — Yossef direct, or delegated by Stark for a social video
  • Workspace~/.openclaw/workspace-phoenix
  • Skillsvideo-use (the studio engine: transcribe→cut fillers/silence→captions→grade→render) · hyperframes(-cli/-media) · elevenlabs-tts (voiceover) · elevenlabs-audio (music) · gdrive (catalog)
  • Engine~/.openclaw/video-use (browser-use/video-use, MIT, isolated venv, Shield-vetted). LLM reads transcript+composites, never frames. Auto: filler/silence removal, EDL cuts, 30ms fades, burned captions, color grade, social formats (9:16 / 16:9).
  • Voice — narrates as Shuri 🐆
  • Deps — pre-approved + pinned at ~/.openclaw/hyperframes-deps (Shield-vetted, —ignore-scripts);
    never auto-installs npm
  • Red lines — drafts for approval only; no auto npm-install; brand-exact feature videos

Drive video pipeline 📁

Phoenix works off gdrive:Phoenix-Video/:

1-Inbox-Raw  →  (Phoenix edits)  →  2-Drafts (v1, v2…)  →  [Yossef approves]  →  3-Production
                                                                              ↘  0-Archive (raw + losing drafts)

Yossef drops raw clips in 1-Inbox-Raw; Phoenix pulls, edits (captions / Shuri voiceover when asked /
background music mixed UNDER existing speech / overlays), and uploads versioned drafts to 2-Drafts; the approved cut lands in 3-Production.
Move-not-delete — finished material parks in 0-Archive.