Your AI assistant generates images for you
Claude Code
U
Generate a hero image for my landing page - a futuristic city at sunset with flying cars
C
I'll start a Masonry image job with FLUX 3 Image, then wait for it and save the file straight into your project.
$ masonry image "futuristic city at sunset, flying cars, cyberpunk aesthetic, cinematic lighting" --model flux-3-image --aspect 16:9
{ "success": true, "job_id": "img_8x7k2m", "status": "pending" }
$ masonry job wait img_8x7k2m --download -o ./public/futuristic-city.png
{ "success": true, "data": { "status": "completed", "file": "./public/futuristic-city.png" } }
The file is now at public/futuristic-city.png. I can reference it in the landing page or run the same prompt through another model for comparison.
Images, video and voice from one CLI
Every command
- image
- Images from text, edits, references, characters
- video
- Text or image to video, first and last frames, edits
- audio
- Voiceover, music, sound effects, transcripts
- production
- Multi-shot videos from a JSON spec, rendered
- character
- Your workspace's reusable people and mascots
- scrape
- Product data and images from a web page
- layers
- Split an image into transparent layers
- guide
- Which model to use for a brief
- models
- Live catalog, prices and per-model flags
- job
- Status, wait, wait-many and download
Run masonry models params <model> for a model's exact flags, or masonry <command> --help for the rest.
Current image and video routes, one CLI
Frequently asked questions
Masonry CLI starts AI image, video and audio jobs from your terminal and downloads finished files into your project. The authenticated catalog returned 36 image routes and 23 video routes when checked on November 5, 2026.
- Can Claude Code generate images?
- Not on its own. Claude Code is a text coding agent with no built-in image model. With the Masonry CLI installed, you can ask it for an image or a video in plain language and it runs the command for you, saving the file into your project. Run masonry skill install once so Claude Code knows the commands.
- How do I generate an image from the terminal?
- Run npm install -g @masonryai/cli, authenticate with masonry login, then run masonry image "your prompt". The command returns a job ID at once. Run masonry job wait <job-id> --download -o ./image.png to wait for it and save the file, or masonry job download <job-id> --output ./image.png later.
- Which AI models does the Masonry CLI support?
- The authenticated catalog returned 36 image routes and 23 video routes when checked on November 5, 2026, including GPT Image 2.5, Nano Banana 2.1, FLUX 3, Seedream 5 Pro, Ideogram V4, Recraft V4, MiniMax H3, Kling 4.0, Veo 3.1, WAN 3.0, and Seedance, plus lip-sync, talking-head and upscaling routes. Omit --model to use Masonry's current default, or run masonry guide for which model suits a brief. Run masonry models list for the live catalog, because it changes.
- Can it make a finished video with voiceover and music?
- Yes. masonry audio speech, music and sfx make the voiceover, soundtrack and effects. masonry production create --spec takes a JSON spec of the edit and uploads the local files it points to, then masonry production render <id> --wait renders it and masonry production download fetches the video.
- Can it keep the same person or mascot across images?
- Yes. Characters you save in your Masonry workspace are listed by masonry character list. Pass one to an image with --character <name>, which sends its sheet and looks as references so the same person stays the same.
- Do I need an API key?
- No raw provider keys. You authenticate once with masonry login, which connects the CLI to your Masonry account, and generations draw from your account balance.
- Does it work with other AI coding assistants, not just Claude Code?
- Yes. The CLI is a normal shell command, so Codex, Cursor or any agent that can run shell commands can call it. To use Masonry inside ChatGPT, Claude or Claude Code without the terminal, connect it at masonry.so/connect.
- Can I use Masonry in Claude chat instead of the terminal?
- Yes. Connect it at masonry.so/connect: the Claude button opens Claude's connector form with Masonry already filled in. Then you can generate, remix, animate, and upscale images and video from a Claude conversation, with your Masonry workspace. Use the CLI for scripted, repeatable jobs and the connector when you work in chat.
- What can I control on a generation?
- Choose a route with --model and request an aspect ratio with --aspect or exact pixels with --dimension. Images take source and reference images (-i, --ref) and characters; videos take a first and last frame (--image, --last-image), a clip to edit (--video), a duration and audio on or off. Inspect any route's inputs with masonry models params <model>, and choose the local file with -o.
Get started in seconds
Need an account? Sign up free