Portal

What is Kaviris?

Kaviris runs the app you built with an AI tool (Claude Code, Cursor, Lovable, or anything else) on the internet, at its own address. You don't need to know about servers. Your coding agent does the work, and you approve what matters in plain language.

It is built for people who make an app with AI and then wonder whether it is safe to put online. Kaviris checks that for you every time you deploy.

What happens when you deploy

  1. Your source is uploaded. Your agent (or the kaviris command) packs the project and sends it. Files like .env are never included.
  2. Kaviris builds it. It works out the language and how to start the app. You don't write a Dockerfile.
  3. Kaviris checks it. It looks for keys and passwords written into the code, and for libraries with known security holes (those it recommends upgrading, with the patch ready, but never blocks). Then it starts the app in a sealed test machine and watches what it tries to connect to.
  4. You approve the contract. On the first deploy you get a short summary like "Talks to api.openai.com (OpenAI). Stores data in Postgres. Needs OPENAI_API_KEY." That summary is the contract: the app can do what it says and nothing else.
  5. It's live at https://<name>.apps.beta.kaviris.com.

If something is wrong, the deploy is stopped with a refusal that says what was found, why it matters and how to fix it. Your agent can usually fix it on its own and deploy again.

What you get

Get started

Pick the tool you build with:

Kaviris is in free beta. See beta limits.

Learn more

This page as Markdown