Share a Lovable or other AI-built app privately

To share an app built with Lovable or another AI app builder privately, take a copy of its code, open it in an AI tool that is connected to Inhaus (such as Claude Code, Cursor or VS Code), and ask it to deploy and share the app. Inhaus puts the app behind your company login, so only the people you add can open it.

Before you start

• A copy of the app's code on your computer. Many AI app builders let you download the code or sync it to a GitHub repository; check your builder's docs for how.
• An AI tool that can read the files on your computer and supports MCP connectors: Claude Code, Cursor, VS Code or Windsurf.
• An Inhaus account and a workspace. Create one at inhaus.dev/getstarted. Your first 7 days are free.

Steps

1. Get the code

   Download or clone the app's code into a folder on your computer.

2. Connect your AI tool to Inhaus

   Add the connector URL to your AI tool and sign in on inhaus.dev:

   Copy

   https://mcp.inhaus.dev/mcp

   For Claude Code, that is one command:

   Copy

   claude mcp add --transport http --scope user inhaus https://mcp.inhaus.dev/mcp

   Then run /mcp, choose inhaus and Authenticate.

3. Open the folder and ask it to deploy
   Copy

   Deploy this project to Inhaus as "Client Portal".

   Your AI tool sends the files. Inhaus installs the packages in package.json, bundles TypeScript and React, runs a security scan and puts the app live. If the project already has a built dist/ folder, sending just those files works too.

4. Fix what comes up

   If the build or the scan finds a problem, your AI tool gets the details and a fix prompt. Ask it: "Fix what Inhaus found and redeploy." See What to check below.

5. Share it privately
   Copy

   Share Client Portal with the account managers team, view only.

   They sign in with their company Google or Microsoft account. Nobody else can open the app.

What to check

Apps from AI builders often lean on services set up for them by the builder. Look for these:

• It has to be a web app in HTML, JavaScript or TypeScript. Most builders make React apps, which work. Python is planned. See Deploy an app.
• Server code runs without Node.js built-ins. Code that needs fs, net or child_process fails to build. Ask your AI tool to use web APIs instead. See Apps with a server.
• Outside services are checked. If the app talks to a hosted database or API, that site must be on your company's allowed domains. A new one is flagged and sent to your admin.
• Keys go in secrets. Move any API key in the code into an Inhaus secret with "store this key as a secret", and read it from server code. The scan blocks keys written into code. See Secrets.
• Drop the app's own login. Inhaus signs people in before the app loads. If the app has its own sign-in screen, ask your AI tool to remove it and read the person from getUser().

Keeping it up to date

The copy on Inhaus does not change when you edit the app in your builder. After you change it there, take the new code and ask your AI tool to deploy again. It becomes a new version at the same link, and everyone keeps their access. Or make changes directly with your AI tool and redeploy.

FAQ

Next steps

• Quickstart
• How to deploy vibe-coded internal tools securely
• Project structure