About CloudShell

Insites CloudShell is the bridge between your AI assistant and your Insites instance. It lets an AI assistant deploy code, upload assets, read logs, and manage settings on your instance, using your own Insites sign-in. You describe what you want in plain language, and your AI does the work through CloudShell.

CloudShell is a remote MCP server. MCP (Model Context Protocol) is the open standard that lets AI assistants use outside tools, so CloudShell works with any AI assistant that supports it, including Claude Code, Claude, Cursor, Windsurf, and ChatGPT.

What CloudShell Does

• Deploys your code - pages, partials, layouts, GraphQL, schema, and every other file in your codebase.
• Uploads assets - images, fonts, and PDFs, returning the public URL to use in your pages.
• Connects to GitHub - commits your AI's work to a repository and deploys on every push, through the Insites GitHub App.
• Shows what happened - deploy status, deployment history, and your instance logs.
• Manages settings - environment variables and the list of installed modules.

Why CloudShell Exists

• Nothing to install, no keys to manage. You do not install command-line tools or store instance API keys on your computer. You sign in with your Insites account, and CloudShell shows every instance you have access to.
• Works from any AI assistant, even on the web. Because CloudShell runs online and signs you in through your browser, it works the same in a desktop coding tool or a web chat. A web-based AI with no access to your files can still deploy a file from a public link.
• One recorded path for every deploy. Whether a change comes from your AI, a direct upload, or a GitHub push, it is recorded in one deployment history for the instance. When a direct, upload, or link deploy replaces a file that is already on the instance, CloudShell keeps a copy of the replaced file for 30 days. This includes a file that reached the instance another way, such as from GitHub, if it was there when CloudShell last read the instance. See Undoing a Change.
• Built for how AI assistants work. Large files are sent straight from your computer to CloudShell rather than being retyped by the AI, and their contents are checked before deploying. This keeps big files from being cut short or corrupted.

How It Works

This is what happens when you ask your AI assistant to build and deploy something.

1. You sign in once. When you connect CloudShell, your browser opens and you sign in with your Insites account. CloudShell then acts as you, with the same access you have in the Console.
2. You choose an instance. Your AI lists the instances in the Insites accounts you belong to, and selects the one you want to work on.
3. Your AI writes the code. Guided by the Insites Logic Engine, it writes files that follow Insites conventions.
4. CloudShell deploys it. Small files are sent directly. Large files and images go through a secure upload, and their contents are checked before they deploy. If you have connected GitHub, your AI commits the files instead and the push deploys them.
5. Your AI checks its work. It reads the deploy status and your instance logs, so it can catch an error and fix it without you having to look.

Ways to Deploy

Connecting CloudShell

Add CloudShell to your AI assistant as a remote MCP server with this address, then sign in with your Insites account:

https://cloudshell.insites.io/mcp

For step-by-step instructions, see Getting Set Up with Insites and AI.

Security

• Your Insites sign-in, nothing else. CloudShell uses the same secure sign-in as the Console. No passwords or API keys are stored in your AI assistant's settings.
• Access follows your role. CloudShell can only reach instances in your Insites accounts, and what it can do follows your role. By default, Viewer can read logs and history, Editor can also set environment variables and push to a branch, and Developer or above can deploy.
• You stay in control. Most AI assistants ask before using a tool unless you allow it, and you can disconnect CloudShell from your AI assistant at any time.