Install an AI Agent for Image and Video Creation with ArtArch CLI

An AI agent image and video creation CLI gives your existing agent a practical way to work with creative projects. ArtArch CLI adds image, video, workflow, and website creation skills to the agent you already use, so you can describe the job in natural language and let the agent operate an ArtArch Studio canvas from the command line.

Start with the agent you already use

Open the ArtArch CLI page and copy the installation instruction shown for AI agents. The instruction installs the ArtArch CLI skill package. After the agent finishes configuring it, restart the agent so the new commands and skill instructions are available in the next session.

The current installation instruction points to the public ArtArch CLI skill package:

Install the ArtArch CLI skill from https://d3ohi5svx56rr5.cloudfront.net/artarchai/artarch-cli/0.1.1/artarch-cli-skill.zip

The page also has a manual CLI option for people who want to install the binary themselves. The agent route is useful when you want the agent to handle the setup and then continue into a creative task.

What the skill gives your agent

The CLI covers the working parts of an ArtArch Studio project. Your agent can inspect available node types, list projects, select a canvas, update nodes, connect compatible inputs, start a node or full flow, wait for a terminal result, and download finished artifacts.

That changes the handoff. You can ask for a square product image, a reference-led video, or a multi-step creative workflow while the agent keeps the project structure in view. The command line remains useful for repeatable checks and downloads, while the agent handles the conversational layer.

Make the first session predictable

Log in through the browser before the agent needs to access Studio:

artarch login web
artarch account info

Then inspect the available projects and choose the canvas for the task:

artarch project list | jq
artarch project use <canvas-id>
artarch node list

If you prefer to keep the directory free of a canvas binding, pass the project ID directly to commands that support --project. Ask the agent to inspect the project before changing it. That gives you a visible starting point and keeps the request tied to one canvas.

Give the agent a concrete creative brief

Short requests work best when they name the input, the visible result, and the format. For example, ask for a product image with a centered object and a square composition, or ask for a short image-to-video scene that keeps the supplied subject in the same setting while the camera moves closer.

The agent can turn that brief into the project configuration, but you still control the run. Generation commands can consume credits, so review the selected project, node, upstream inputs, and settings before confirming a run. After a run starts, use its run ID for waiting or downloading instead of submitting a duplicate request.

Frequently asked questions

Which AI agent can use ArtArch CLI?

The ArtArch CLI page provides an AI-agent installation flow for the agent you already use. Install the skill package, restart the agent, and then ask it to create with ArtArch.

Does installation create a project automatically?

The first useful session starts by listing or selecting an existing Studio project. You can then inspect its nodes and configure the canvas for the image or video task.

How does an agent deliver the finished asset?

After a node or flow reaches a terminal state, the CLI exposes the run result and can download available artifacts locally with the run download command.

What should I check before generation?

Check the canvas, the selected node, upstream dependencies, connected references, and the intended output before approving the run. This keeps the agent's action within the creative scope you asked for.

To try the setup, open the ArtArch CLI page, choose the AI-agent installation flow, and ask your agent to create with ArtArch.