CLI v1.0

Pipeline CLI Reference

AI content generation, video analysis, transcription, YAML pipelines, and agentic video production — all from the command line.

$ qcut <group> <action> [options]

Installation

The CLI is bundled with QCut. Commands use a <group> <action> syntax:

# Group syntax (recommended)

qcut <group> <action> [options]

# Groups: gen, analyze, audio, edit, flow, system

qcut gen image -t "A cat in space"

# Production binary

qcut <group> <action> [options]

Quick Start

Every qcut gen, qcut flow, and qcut analyze command deducts credits — so you have to log in before running any of them. Follow Step 1 first, then come back to Step 2.

1. Log in

Every AI command deducts credits from your account. Sign up once (free tier includes 50 credits/month), then log in from the CLI.

# Google OAuth (easiest — opens a browser window)

qcut system login

# Email + password

qcut system login --email you@example.com

# Scripted: pipe credentials via env vars (no interactive prompt)

qcut system login --email "$QCUT_EMAIL" --password "$QCUT_PASSWORD"

Beta testers: use the @qcut.app credentials emailed to you. Each test account is pre-loaded with 1000 credits on the free plan.

export QCUT_TEST_EMAIL=test@qcut.app
export QCUT_TEST_PASSWORD='...'
qcut system login --email "$QCUT_TEST_EMAIL" --password "$QCUT_TEST_PASSWORD"

Verify with qcut system check-keys --json — a successful login shows your user email and remaining credits.

Check balance directly against the license server:

curl -H "Authorization: Bearer $(qcut system get-key --name QCUT_AUTH_TOKEN --reveal)" https://qcut-license-server.zdhpeter.workers.dev/api/credits/balance

2. Run a command

Each of these requires an active session — if you skipped Step 1 the CLI will exit with API key missing (exit code 4).

# Generate an image

qcut gen image -t "A cinematic portrait at golden hour"

# Create a video

qcut gen video -m kling_2_6_pro -t "Ocean waves at sunset" -d 5s

# Transcribe audio

qcut analyze transcribe -i audio.mp3 --srt

# Run a YAML pipeline

qcut flow run -c pipeline.yaml -i "A sunset" --no-confirm

# Full video from an idea (ViMax)

qcut flow idea2video --idea "A detective in 1920s Paris" -d 120

3. Log out when done

Clears the stored QCUT_AUTH_TOKEN from ~/.qcut/.env.

qcut system logout

Progressive Help System

3-level structured help available with --help --json.

# Level 1: All groups and commands

qcut --help --json

# Level 2: Command detail

qcut gen image --help --json

# Level 3: Parameter detail

qcut gen image --help model --json