Creator Edit: camera + clean mic → finished short

Rules: ${CLAUDE_PLUGIN_ROOT}/references/rules.md · CLI: ${CLAUDE_PLUGIN_ROOT}/references/cli.md. ss = "${CLAUDE_PLUGIN_ROOT}/bin/shorts-studio". Input: $ARGUMENTS

This repeated-take policy is for creator recordings only — never apply it to podcasts/interviews.

0. Engine check

The commands below run the Shorts Studio engine. Installed as the Claude Code plugin it is already present and shorts-studio is on the Bash PATH. If shorts-studio doctor is not found (for example, this skill was added with npx skills add), ask the user before installing anything, then:

• Claude Code: /plugin marketplace add bloxy-studios/shorts-studio → /plugin install shorts-studio@shorts-studio.
• Other agents: git clone --depth 1 https://github.com/bloxy-studios/shorts-studio ~/.shorts-studio/engine and read every ${CLAUDE_PLUGIN_ROOT} path below as ~/.shorts-studio/engine.

Then run shorts-studio doctor --setup once (project-local runtime; never sudo or global installs).

Steps

1. Project (ask only what is missing; autonomous → luxury-tech if the user asked for it, else social-dynamic): ss project init "<name>" --preset <id> --mode <guided|director|autonomous>
2. Plan the edit: ss creator <project> --camera <file> --mic <wav> [--preset luxury-tech] [--title "..."] [--brief graphics.json]. This ingests both files, transcribes the clean mic, analyzes the camera, aligns, detects takes, measures line endings (10 ms RMS: −35 dBFS cue, sustained −32 dBFS for 150 ms) and assembles.
3. Read clips/<clip>/creator-report.md and verify with the user (guided) or log (autonomous):
   • Alignment: confirmed needs agreeing windows and spot checks; probable (transcript fallback) must be stated as such. Never call sync confirmed without that evidence.
   • Each take choice and what a rejected last take said; removed false starts and chatter.
   • Deviations (endpoints extended for quiet consonants, black camera spans needing a visual).
4. Luxury Tech: continue with the luxury-tech skill (brief → plan → preview). Other presets: ss render.
5. Render + QA + inspect: ss render <project> --clip creator-edit; read the contact and cut sheets; watch for lip sync after J-cuts at every join (cut sheet pairs). ss creator … --render also copies the final to edit/final.mp4.

Error handling and limits

• Sync unreliable and too few matching phrases → stop and say the files may not be from the same session.
• A take swap is only for audible word-level defects (stutter, restart, cut-off word, low confidence, ≥ 15 % duration anomaly); parallel sentences ("cars drive themselves / planes fly themselves") are separate lines.
• whisper.cpp timestamps can drift at a chunk end; words overrunning measured speech are re-timed inside it and flagged.
• Synthetic test voices do not match lips; real recordings must be lip-checked on rendered frames.

Example

"Here is camera.mp4 and mic.wav. Use Luxury Tech, title it from the hook, do everything yourself." → ss project init "launch" --preset luxury-tech --mode autonomous; ss creator launch --camera camera.mp4 --mic mic.wav --preset luxury-tech; luxury-tech skill; ss render; ss repair; ss export.

Completion check

The creator report shows a confirmed (or explicitly qualified) alignment, every line's take choice with reasons, removed material; the render's QA has no hard failure; timing.end passes (END = last voice sample rounded to a frame, no frozen/silent tail); you inspected the cut sheet for lip sync.