IMA Studio CLI

Command-line tool for IMA Studio — AI content generation platform.

Generate images, videos, music, and speech from the command line. Designed for both human use and AI agent integration.

Install

npm install -g imastudio-cli

Requires Node.js 18+.

Quick Start

# 1. Configure your API key
ima init

# 2. See what you can create
ima list-task-types

# 3. List available models
ima list-models text_to_image

# 4. Generate an image
ima create-task \
  --task-type text_to_image \
  --model-id doubao-seedream-4.5 \
  --prompt "a mountain sunset, photorealistic" \
  --wait

# 5. Check a task
ima task-status <task-id>

Task Types

| Task Type | Input | Output | Description | |-----------|-------|--------|-------------| | text_to_image | prompt | image | Generate an image from a text prompt. | | image_to_image | prompt + image(s) | image | Edit a single image (style transfer, object removal, etc.) or fuse multiple reference images into one. | | text_to_video | prompt | video | Generate a video from a text prompt. | | image_to_video | prompt + image | video | Generate a video using the input image as the first frame. | | first_last_frame_to_video | prompt + 2 images | video | Generate a video that starts with the first image and ends with the second. | | reference_image_to_video | prompt + image(s) | video | Generate a video using reference images for visual guidance (characters, style, objects) — not as literal frames. | | text_to_music | prompt | audio | Generate instrumental music from a description, or a full song (vocals + music) from lyrics. | | text_to_speech | text | audio | Convert text into spoken audio (TTS). |

Commands

| Command | Description | |---------|-------------| | ima init | Interactive setup — configure API key | | ima list-task-types | List supported task types | | ima list-models <type> | List models for a task type | | ima model-info <model-id> | Detailed model info (pricing, params) | | ima upload <file> | Upload a file to IMA CDN | | ima create-task | Create a generation task | | ima task-status <id> | Check task status / get results | | ima doctor | Diagnose environment and connectivity |

API Key

Your IMA API key is read from (in priority order):

1. --api-key CLI flag
2. IMA_API_KEY environment variable
3. ~/.imastudio/credentials file

Get your key at imastudio.com.

# Option 1: Interactive setup (recommended)
ima init

# Option 2: Environment variable
export IMA_API_KEY=ima_your_key_here

# Option 3: Per-command
ima list-models text_to_image --api-key ima_your_key_here

Examples

Text to Image

# SeeDream 4.5 (default, 5 pts)
ima create-task \
  --task-type text_to_image \
  --model-id doubao-seedream-4.5 \
  --prompt "cute cat wearing a tiny hat" \
  --wait

# Nano Banana Pro, 4K resolution
ima create-task \
  --task-type text_to_image \
  --model-id gemini-3-pro-image \
  --prompt "product photo of a coffee cup" \
  --size 4K \
  --wait

# 16:9 widescreen
ima create-task \
  --task-type text_to_image \
  --model-id doubao-seedream-4.5 \
  --prompt "panoramic city skyline at dusk" \
  --aspect-ratio 16:9 \
  --wait

Image to Image

# Style transfer
ima create-task \
  --task-type image_to_image \
  --model-id doubao-seedream-4.5 \
  --prompt "turn into oil painting style" \
  --input-images ./photo.jpg \
  --wait

# Using a URL as input
ima create-task \
  --task-type image_to_image \
  --model-id gemini-3-pro-image \
  --prompt "remove background, keep subject" \
  --input-images https://example.com/photo.jpg \
  --wait

JSON Output (for scripts & agents)

# All commands support --json
ima list-models text_to_image --json
ima create-task --task-type text_to_image --model-id doubao-seedream-4.5 --prompt "hello" --wait --json

Upload Files

# Upload an image and get CDN URL
ima upload ./my-photo.jpg
# URL: https://ima-ga.esxscloud.com/...

Agent Integration

This CLI is designed to be the single tool an AI agent needs to interact with IMA Studio. The agent workflow:

ima list-task-types          → discover capabilities
ima list-models <type>       → pick a model
ima model-info <id>          → get pricing/params
ima upload <file>            → upload inputs (if needed)
ima create-task ... --wait   → generate content
ima task-status <id>         → check async results

All commands support --json for machine-readable output. No hardcoded model tables needed — everything is fetched at runtime.

Configuration

~/.imastudio/
├── credentials     # API key (chmod 600)
├── config.json     # User preferences (future)
└── cache/          # Cached product lists (1h TTL)

Available Models (as of 2026-04)

Image Generation

| Model | ID | Cost | Notes | |-------|----|------|-------| | SeeDream 4.5 | doubao-seedream-4.5 | 5 pts | Default, 4K, aspect ratio support | | Nano Banana 2 | gemini-3.1-flash-image | 4-13 pts | Budget option, size tiers | | Nano Banana Pro | gemini-3-pro-image | 10-18 pts | Premium, 1K/2K/4K | | Midjourney | midjourney | 8-10 pts | Artistic styles |

Run ima list-models <type> for the latest — models and pricing update without CLI version changes.

License

MIT