video-editing

The ClipReady CLI — agents-first AI video editing for talking-head reels. An AI agent (or you) drops in a raw recording; it is imported to the ClipReady cloud, which transcribes and screens it and sends back the editorial brief; the agent authors a timeline.json (cuts, captions, zooms, overlays); the cloud compiles, renders, and verifies it.

Zero setup — ffmpeg is bundled with the skill. Nothing to install beyond Node 20 and a ClipReady account: import prepares the source locally with the bundled binaries (or lets the cloud ingest it), and every other step — transcription, compile, render, verify, QA — runs in the ClipReady cloud.

Requirements

• Node ≥ 20
• A ClipReady account — the CLI opens a browser to sign in

That's the whole list.

Install & auth

npm install -g video-editing
video-editing auth login              # browser sign-in once per machine
video-editing --help

The base URL defaults to the hosted ClipReady cloud. CLIPREADY_TOKEN / CLIPREADY_API_BASE env vars (or --token/--api-base) override when set. CLIPREADY_API_KEY and --api-key remain deprecated aliases for one release; old API keys themselves are not accepted.

Every cloud command requires the saved session — nothing runs without it.

The companion agent skill (workflow + timeline vocabulary for AI agents):

npx skills add ItayElgazar/clipready --skill video-editing
# or, after installing the CLI:
video-editing skill install

Workflow

video-editing import session                               # mint token + endpoint
node <asset-import-skill-dir>/scripts/upload-media.mjs \
  --token <token> --endpoint <endpoint> \
  --job-root "$HOME/clipready-jobs" "clip.mp4"              # use returned job_dir
# Fallback when no local ffmpeg is usable:
video-editing init --source "clip.mp4" --job-dir jobs/clip
# Continue below with the helper's returned job_dir (or the init job dir).
JOB_DIR="<returned job_dir>"
# read "$JOB_DIR/prompts/author-edl.prompt.md", then author "$JOB_DIR/timeline.json"
video-editing timeline compile --job-dir "$JOB_DIR" --json  # cloud compile → resolved/plan.json + watch link
video-editing files push --job-dir "$JOB_DIR" timeline.json # exact authored bytes become durable VFS truth
video-editing render --job-dir "$JOB_DIR"                   # cloud preview render (waits + downloads)
video-editing verify --job-dir "$JOB_DIR" --json            # cloud content-verify of the preview
video-editing qa frames --join 1 --job-dir "$JOB_DIR"       # server-rendered QA filmstrips of join 1
video-editing render --mode final --job-dir "$JOB_DIR"

Commands

| Verb | What it does | | --- | --- | | import session | Mint a short-lived import token for the bundled client-side import helper — probes/conforms the source with bundled ffmpeg and uploads the transcription audio first, so transcription runs during the video upload | | init --source <path> | Server-ingest import: stream the source up (chunked, 2GB-safe), create the cloud job, run server-side ingest (probe + transcription + screening), pull transcripts/brief — the fallback when no local ffmpeg is usable | | transcribe | Re-run/pull transcription (init already did it once) | | timeline validate | Schema/semantic check of your timeline.json (local, pure) | | timeline resolve | Phrase → exact source seconds (anchor resolution) | | timeline compile | Cloud-compile timeline.json → resolved/plan.json + diagnostics; prints the watch link | | render | Cloud render of the compiled plan (preview by default, waits + downloads; --mode final for delivery) | | verify | Fully server-side content-verify of the rendered preview (retakes left in, dead air) | | qa frames / qa waveform | Server-rendered filmstrip / waveform PNG + word-timing sidecar; address by source time, plan join (--join <i>), or output time (--out-start/--out-end) | | files pull / files push | Sync job artifacts with the cloud job | | review pull / review push | Work a ClipReady review round (live timeline updates, completion) | | cloud render / cloud status | Same render API, legacy spelling + status polling | | skill install | Install the bundled agent skill into ~/.claude/skills | | status | Job state + artifact existence | | config | Store the session token/base in ~/.config/video-editing/.env |

Run video-editing <cmd> --help for flags.

Links

• ClipReady + this skill: https://github.com/ItayElgazar/clipready
• Skill install: npx skills add ItayElgazar/clipready --skill video-editing