@lavendly/cli

Drive agentic AI video generation and editing from the terminal, against the same /v1 REST surface the SDKs, MCP servers, and canvas use.

npx @lavendly/cli --help
# or
npm i -g @lavendly/cli

Auth

Mint a key in the app (Settings → API), then:

lavendly login lv_xxxxxxxx          # saved to ~/.lavendly/config.json
# or
export LAVENDLY_API_KEY=lv_xxxxxxxx

Point it at a stack with LAVENDLY_API_BASE (or --base):

export LAVENDLY_API_BASE=http://localhost:5173/api   # local docker stack
export LAVENDLY_API_BASE=https://lavendly.ai/api      # hosted (default)

The key is sent as Authorization: Bearer <key>, so spend is attributed to it.

Commands

| Command | What it does | |---|---| | lavendly whoami | Active base + key, plus /v1/_status | | lavendly video "<prompt>" [--model auto --duration 5 --image <url> --wait] | Generate a clip; --wait polls for the result | | lavendly tts "<text>" [--voice <id> --format mp3] | Text to speech | | lavendly music "<prompt>" [--duration 30] | Background music | | lavendly chat "<prompt>" [--model <id>] | One-shot completion (scripts, ideation) | | lavendly workflows / workflow <id> | List / get workflows | | lavendly render <wf_id> [--wait] | Render a workflow to a final video | | lavendly render-status <wf_id> <job> | Poll a render job | | lavendly usage | Monthly usage / credits | | lavendly schema | Machine-readable API contract |

Global flags: --json (compact output), --key, --base, --help.

Example: prompt to clip

lavendly login lv_xxxx
lavendly video "a neon koi swimming through rain, 35mm, shallow depth of field" \
  --duration 5 --wait

Zero dependencies (Node 18+ global fetch).