ai-generator

Node.js CLI for AI Generator image and video generation from a terminal, script, or personal Agent.

Requirements

• Node.js 18 or newer
• The production server URL: https://aiaigenerator.ai (override with AI_GENERATOR_BASE_URL only for development or self-hosted deployments)
• A CLI access key created from the server's /cli page

Install

npm install --global @ai-generator/cli

For local development from this repository:

node cli/bin/ai-generator.mjs --help

Authentication

Prefer an environment variable for Agents, CI, and automation:

export AI_GENERATOR_API_KEY="ai_live_..."
export AI_GENERATOR_BASE_URL="https://aiaigenerator.ai"

For local Agent setup, save the key on the device in a plain-text file first. Do not send the key in the Agent chat:

mkdir -p ~/.ai-generator
${EDITOR:-nano} ~/.ai-generator/cli-key.txt

Then let the CLI read that local file and write its protected configuration at ~/.ai-generator/config.json:

ai-generator auth set --key-file ~/.ai-generator/cli-key.txt --base-url "https://aiaigenerator.ai"
ai-generator auth show --json

On Windows, use %USERPROFILE%\.ai-generator\cli-key.txt as the local file path. The --key-file option reads the key without putting it in command arguments or CLI output. Keep the file out of Git and shared folders; delete it after setup if the local config is sufficient.

For CI or other non-interactive automation, use AI_GENERATOR_API_KEY through the secret manager instead of a text file.

Commands

ai-generator generate-image --prompt "A product photo" --wait --json
ai-generator generate-image --prompt "Edit this image" --image ./reference.png --wait --json
ai-generator generate-image --prompt "A 4K product photo" --model google/nano-banana-pro --size 16:9 --resolution 4K --wait --json
ai-generator generate-video --prompt "A slow product reveal" --aspect-ratio landscape --frames 10 --wait --json
ai-generator task-status --task-id <id> --type video --json
ai-generator task-download --task-id <id> --type video --output-dir ./outputs --json
ai-generator credits --json

generate-image and generate-video call the image and video generation APIs directly. The CLI does not use the website's internal Agent drawing workflow. Image requests support --model, --size, --resolution, and --number-of-images; the server calculates the final credit cost from the selected model and parameters instead of trusting a client-supplied cost. Video supports text-to-video or image-to-video based on whether --image is provided.

Use --model to select a configured image or video model. The supported model and parameter combinations are validated by the CLI and server. Use --help and --version to inspect the local command surface.

JSON output

Use --json when another Agent or script consumes the result. Creation returns a task ID. With --wait, the response also includes the normalized terminal task status. Result files can be downloaded after a successful task.

Development

pnpm cli:test
npm pack --dry-run