Claude Skill

editor-handoff

Editor hand-off for Claude: packages a Video Studio edit so another editor can finish it in Premiere, DaVinci Resolve, Final Cut or Avid: CMX3600 EDL, FCPXML 1.10 and OpenTimelineIO timelines with markers, the graphics layer as ProRes 4444 with alpha, a ProRes or DNxHR picture re

LLM Mart · 0 points · 0 views 0 listing impressions 0 install-command copies
Download viprasol-tech-claude-video-editing-skills-plugins_video-studio_skills_editor-handoff-0293497.zip · 3 KB

Install

skills CLI npx skills add https://github.com/Viprasol-Tech/claude-video-editing-skills/tree/main/plugins/video-studio/skills/editor-handoff
Claude Code claude plugin marketplace add https://llmmart.ai/marketplace.json && claude plugin install viprasol-tech-claude-video-editing-skills@llmmart
Git git clone https://github.com/Viprasol-Tech/claude-video-editing-skills.git

The skills CLI installs just this skill, for any of its supported agents. Claude Code installs the whole viprasol-tech/claude-video-editing-skills collection as a plugin from our marketplace. Git is the plain clone.

Skill manifest

Editor Handoff

A turnover a stranger can open at 2 a.m. and understand: timeline, media, graphics with alpha, mix, captions, notes.

  • Best for: finishing in an NLE, colourist or mixer turnovers, agency hand-offs, archiving an edit
  • Inputs: a Video Studio work folder W (after run.py/autocut.py, ideally rendered), the receiving NLE
  • Outputs: W/handoff/ with EDL, FCPXML, OTIO, front_4444.mov, source.mp4, picture reference, mix WAV, captions.srt, README.md

Before you start

  • This skill needs video-studio-core in the same skills folder. CORE = that folder.
  • First run on a machine: python CORE/scripts/check_env.py. opentimelineio is optional (OTIO is written by hand without it).
  • Ask: which NLE and version, frame rate the project must use, timecode start (default 01:00:00:00), and whether they want burned graphics or separate alpha layers. Picture lock first: a hand-off of a moving edit wastes everyone's day.

What gets handed over

file made by what it is
<name>.edl export_edl.py CMX3600, one video track, reel = file stem, * FROM CLIP NAME, * LOC markers
<name>.fcpxml export_edl.py FCPXML 1.10 (Final Cut; Resolve and Premiere import it), asset-clips on the spine, markers
<name>.otio export_edl.py OpenTimelineIO JSON (Resolve, Premiere via plug-in, Avid via adapters)
source.mp4 the work folder the media the timelines point at (30 fps, upright, normalised by prep.py)
front_4444.mov render.py --keep + export_edl.py --encode-alpha captions and graphics with alpha, full length
final_prores-422hq.mov deliver.py picture reference / mezzanine with the full mix
final_audio-wav.wav deliver.py the mix, 48 kHz 24-bit, untouched loudness
captions.srt/.vtt/.ttml deliver.py subtitles from the transcript
README.md you (Claude) how to conform, what is where, known issues

Limits to say out loud: the timelines carry the cut only (one video track from cutlist.json). Overlays, zooms, colour and the mix are not editable in the NLE: they come as the alpha layer, markers and baked files. Multi-clip timelines from timeline.py export the assembled cut against source.mp4 (= the assembled video), not the original camera files. There are no separate DX/MX/FX stems: hand over the mix WAV, and the music file and voice separately if the mixer needs them.

Make the hand-off

# 1. timelines (record starts at 01:00:00:00; drop-frame used for 29.97/59.94)
python CORE/scripts/export_edl.py W --plan W/plan.json --out-dir W/handoff --record-start 01:00:00:00 --title "client_project_v003"
# 2. graphics layer with alpha: render once keeping the PNG frames, then encode ProRes 4444
python CORE/scripts/render.py W/plan.json --keep
python CORE/scripts/export_edl.py --encode-alpha "W/render/front/%05d.png" --fps 30 --out W/handoff/front_4444.mov
# 3. picture reference + mix + captions
python CORE/scripts/finish.py W/plan.json
python CORE/scripts/deliver.py W/plan.json --to prores-422hq,audio-wav --out-dir W/handoff --burn no --no-thumbs --no-post
# 4. QC the reference
python CORE/scripts/qc.py W/handoff/final_prores-422hq.mov --plan W/plan.json --platform none

render.py --keep leaves one PNG folder per layer in W/render/ (front always when there are overlays; bg and behind only when the plan uses them). Encode each layer you hand over (e.g. behind for text behind the speaker: --encode-alpha "W/render/behind/%05d.png" --out W/handoff/behind_4444.mov; it only works over the matte, so also hand over W/matte.mp4). Then copy W/source.mp4 into W/handoff/. python CORE/scripts/export_edl.py --alpha prints the same recipe.

ProRes 4444 decodes as yuva444p12le in FFmpeg: that is normal. Check alpha by importing it over a colour matte.

Write the README

Write W/handoff/README.md for the receiving editor (plain text, no jargon they would not use):

PROJECT  client_project_v003   FPS 30 (constant)   RES 1920x1080   TC start 01:00:00:00
TIMELINE  client_project_v003.fcpxml (Final Cut / Resolve), .edl (Avid/Premiere), .otio
MEDIA     source.mp4  <- relink to this; reel name = "source"
LAYERS    V1 source (cut per timeline)  V2 front_4444.mov (graphics + captions, alpha, starts at 01:00:00:00)
AUDIO     final_audio-wav.wav = full mix (dialogue chain, music ducked, -14 LUFS when finished for web)
          the timeline audio is the raw camera sound; mute it if you lay the mix
REFERENCE final_prores-422hq.mov = how it should look and sound
CAPTIONS  captions.srt (sidecar), burned captions are inside front_4444.mov
MARKERS   text overlays and zoom punches are markers on the timeline (names = overlay text)
COLOUR    look "<name>" at <intensity> baked into the reference only; source.mp4 is ungraded
KNOWN     <anything unresolved: a name to confirm, a music licence, a shot to replace>
CONTACT   <who to ask>

Check before delivering

  • Import the FCPXML or EDL yourself into any free NLE if available (Resolve), or at least open the EDL as text: event count matches the number of segments in W/cutlist.json, record total equals the length of cut.mp4.
  • ffprobe front_4444.mov: codec prores (4444 profile), pix_fmt with alpha, duration equals the plan duration.
  • Reference and mix durations match the timeline to the frame; qc.py passes.
  • README lists every file, the frame rate, timecode start and known issues. Zip or folder name carries the version (client_project_v003_handoff).

Example requests

  • "My editor wants to finish this in Premiere, give her an XML and the graphics separately."
  • "Export the cut as an EDL with markers for the colourist, start at 10:00:00:00."
  • "Package everything for Resolve: timeline, captions, mix, and a README."
Files (claude-video-editing-skills)
  • SKILL.md 6.4 KB
    ---
    name: editor-handoff
    description: "Editor hand-off for Claude: packages a Video Studio edit so another editor can finish it in Premiere, DaVinci Resolve, Final Cut or Avid: CMX3600 EDL, FCPXML 1.10 and OpenTimelineIO timelines with markers, the graphics layer as ProRes 4444 with alpha, a ProRes or DNxHR picture reference, the mix as WAV, captions as SRT, and a README that explains it all. Use when the user says 'send this to my editor', 'export an EDL', 'XML for Premiere', 'FCPXML', 'OTIO', 'turnover to Resolve', 'give me the graphics with transparency', 'finish it in DaVinci', 'conform this'. Runs on video-studio-core (export_edl, render --keep, deliver, qc)."
    license: MIT
    ---
    
    # Editor Handoff
    
    A turnover a stranger can open at 2 a.m. and understand: timeline, media, graphics with alpha, mix, captions, notes.
    
    - **Best for:** finishing in an NLE, colourist or mixer turnovers, agency hand-offs, archiving an edit
    - **Inputs:** a Video Studio work folder `W` (after `run.py`/`autocut.py`, ideally rendered), the receiving NLE
    - **Outputs:** `W/handoff/` with EDL, FCPXML, OTIO, `front_4444.mov`, `source.mp4`, picture reference, mix WAV,
      `captions.srt`, `README.md`
    
    ## Before you start
    
    - This skill needs **video-studio-core** in the same skills folder. `CORE` = that folder.
    - First run on a machine: `python CORE/scripts/check_env.py`. `opentimelineio` is optional (OTIO is written by hand
      without it).
    - Ask: which NLE and version, frame rate the project must use, timecode start (default `01:00:00:00`), and whether
      they want burned graphics or separate alpha layers. Picture lock first: a hand-off of a moving edit wastes everyone's day.
    
    ## What gets handed over
    
    | file | made by | what it is |
    | --- | --- | --- |
    | `<name>.edl` | `export_edl.py` | CMX3600, one video track, reel = file stem, `* FROM CLIP NAME`, `* LOC` markers |
    | `<name>.fcpxml` | `export_edl.py` | FCPXML 1.10 (Final Cut; Resolve and Premiere import it), asset-clips on the spine, markers |
    | `<name>.otio` | `export_edl.py` | OpenTimelineIO JSON (Resolve, Premiere via plug-in, Avid via adapters) |
    | `source.mp4` | the work folder | the media the timelines point at (30 fps, upright, normalised by `prep.py`) |
    | `front_4444.mov` | `render.py --keep` + `export_edl.py --encode-alpha` | captions and graphics with alpha, full length |
    | `final_prores-422hq.mov` | `deliver.py` | picture reference / mezzanine with the full mix |
    | `final_audio-wav.wav` | `deliver.py` | the mix, 48 kHz 24-bit, untouched loudness |
    | `captions.srt/.vtt/.ttml` | `deliver.py` | subtitles from the transcript |
    | `README.md` | you (Claude) | how to conform, what is where, known issues |
    
    Limits to say out loud: the timelines carry the **cut only** (one video track from `cutlist.json`). Overlays, zooms,
    colour and the mix are not editable in the NLE: they come as the alpha layer, markers and baked files. Multi-clip
    timelines from `timeline.py` export the assembled cut against `source.mp4` (= the assembled video), not the original
    camera files. There are no separate DX/MX/FX stems: hand over the mix WAV, and the music file and voice separately if
    the mixer needs them.
    
    ## Make the hand-off
    
    ```bash
    # 1. timelines (record starts at 01:00:00:00; drop-frame used for 29.97/59.94)
    python CORE/scripts/export_edl.py W --plan W/plan.json --out-dir W/handoff --record-start 01:00:00:00 --title "client_project_v003"
    # 2. graphics layer with alpha: render once keeping the PNG frames, then encode ProRes 4444
    python CORE/scripts/render.py W/plan.json --keep
    python CORE/scripts/export_edl.py --encode-alpha "W/render/front/%05d.png" --fps 30 --out W/handoff/front_4444.mov
    # 3. picture reference + mix + captions
    python CORE/scripts/finish.py W/plan.json
    python CORE/scripts/deliver.py W/plan.json --to prores-422hq,audio-wav --out-dir W/handoff --burn no --no-thumbs --no-post
    # 4. QC the reference
    python CORE/scripts/qc.py W/handoff/final_prores-422hq.mov --plan W/plan.json --platform none
    ```
    
    `render.py --keep` leaves one PNG folder per layer in `W/render/` (`front` always when there are overlays; `bg` and
    `behind` only when the plan uses them). Encode each layer you hand over (e.g. `behind` for text behind the speaker:
    `--encode-alpha "W/render/behind/%05d.png" --out W/handoff/behind_4444.mov`; it only works over the matte, so also
    hand over `W/matte.mp4`). Then copy `W/source.mp4` into `W/handoff/`. `python CORE/scripts/export_edl.py --alpha`
    prints the same recipe.
    
    ProRes 4444 decodes as `yuva444p12le` in FFmpeg: that is normal. Check alpha by importing it over a colour matte.
    
    ## Write the README
    
    Write `W/handoff/README.md` for the receiving editor (plain text, no jargon they would not use):
    
    ```text
    PROJECT  client_project_v003   FPS 30 (constant)   RES 1920x1080   TC start 01:00:00:00
    TIMELINE  client_project_v003.fcpxml (Final Cut / Resolve), .edl (Avid/Premiere), .otio
    MEDIA     source.mp4  <- relink to this; reel name = "source"
    LAYERS    V1 source (cut per timeline)  V2 front_4444.mov (graphics + captions, alpha, starts at 01:00:00:00)
    AUDIO     final_audio-wav.wav = full mix (dialogue chain, music ducked, -14 LUFS when finished for web)
              the timeline audio is the raw camera sound; mute it if you lay the mix
    REFERENCE final_prores-422hq.mov = how it should look and sound
    CAPTIONS  captions.srt (sidecar), burned captions are inside front_4444.mov
    MARKERS   text overlays and zoom punches are markers on the timeline (names = overlay text)
    COLOUR    look "<name>" at <intensity> baked into the reference only; source.mp4 is ungraded
    KNOWN     <anything unresolved: a name to confirm, a music licence, a shot to replace>
    CONTACT   <who to ask>
    ```
    
    ## Check before delivering
    
    - Import the FCPXML or EDL yourself into any free NLE if available (Resolve), or at least open the EDL as text:
      event count matches the number of segments in `W/cutlist.json`, record total equals the length of `cut.mp4`.
    - `ffprobe front_4444.mov`: codec `prores` (4444 profile), pix_fmt with alpha, duration equals the plan duration.
    - Reference and mix durations match the timeline to the frame; `qc.py` passes.
    - README lists every file, the frame rate, timecode start and known issues. Zip or folder name carries the version
      (`client_project_v003_handoff`).
    
    ## Example requests
    
    - "My editor wants to finish this in Premiere, give her an XML and the graphics separately."
    - "Export the cut as an EDL with markers for the colourist, start at 10:00:00:00."
    - "Package everything for Resolve: timeline, captions, mix, and a README."
    

Comments (0)

Sign in to join the conversation.

No comments yet.

Reviews (0)

No reviews yet.