How Varity Works

Varity Team Core Contributors Updated May 2026

Varity deploys Node and Python apps, AI agents, and LLM workloads with a fast, infrastructure-abstracted workflow. You do not need Docker, infrastructure knowledge, or DevOps experience.

Who Varity is for

Varity works seamlessly with AI coding tools to build and ship apps. People reach for it in a few ways:

Building something new with an AI tool (vibe coding): You are using Claude Code, Cursor, or a similar tool to build an app from scratch. You want to go from “working locally” to “live URL” without leaving your editor. Varity’s MCP server handles the entire deploy in one prompt.

Migrating from Vercel: You have an existing app on Vercel and want lower costs. Run varitykit migrate to remove Vercel-specific config, then deploy. Most migrations take under 5 minutes.

Deploying your current project: You already have app code locally or in a GitHub repo and want a live URL fast. Run varitykit app deploy (or ask your MCP client to run varity_deploy) and ship.

In all three cases, you never configure infrastructure. You just build and deploy.

The platform

Varity has three core websites:

Site	What it does
varity.so	Marketing site
docs.varity.so	Documentation and guides
developer.store.varity.so	Developer portal: deploy, monitor, billing

The developer flow

Here is how you go from idea to a live URL:

Set up the MCP server

Add Varity to your AI coding tool once. After that, every project you work on has Varity available.
Terminal window
```
claude mcp add varity -- npx -y @varity-labs/mcp@beta
```
Open your project

Open any existing project or create a new one. You do not need to change any code.
Deploy

Ask your AI tool to deploy:
```
Deploy my app to Varity
```
Or use the CLI:
Terminal window
```
varitykit app deploy
```
Your app is live

Your app is now available at a public URL:
```
Static hosting: https://varity.app/your-app-name/
Dynamic hosting: https://your-app-name.varity.app/
```
Varity detects your framework, builds your project, provisions the resources it needs, and returns a working URL. All of this happens automatically.

What Varity detects automatically

When you deploy, Varity inspects your project and makes all infrastructure decisions:

Framework detection: Varity reads your package.json, requirements.txt, or pyproject.toml to identify your framework (Next.js, React, Vue, FastAPI, Django, etc.) and selects the right build process.

Static vs. dynamic hosting: Projects with a server or backend run in containers. Static sites (React, Vue, Next.js with export mode) go to a global CDN. You can override this with --hosting dynamic or --hosting static if needed.

Auto-wired services: If your project lists database or caching dependencies, Varity provisions those services automatically and injects the connection details at runtime.

Dependency in your project	Service added automatically
`pg`, `@prisma/client`, `drizzle-orm`, `psycopg`	Postgres database
`ioredis`, `redis`	Redis cache
`mongoose`, `mongodb`, `pymongo`	MongoDB
`@langchain/ollama`, `ollama`	Ollama AI runtime

You do not write connection strings or provision databases. Varity handles it.

Environment variables: Varity reads your .env.varity, .env.local, or .env file and passes those variables to your running app. Platform-specific variables from Vercel, Netlify, Render, and similar hosts are filtered out automatically.

Pricing

Varity pricing is based on reserved hardware profiles, while Vercel, Render, and Railway are primarily usage-metered. Your bill is tied to your selected profile and does not change with traffic spikes, bandwidth, or invocation count.

Use the varity_cost_calculator MCP tool and /resources/pricing to estimate workload cost before deploying.

The MCP server

The Varity MCP server gives your AI coding tool a full set of tools for deploying, monitoring, and managing your apps:

Tool	What it does
`varity_doctor`	Check if your environment is ready
`varity_login`	Authenticate with your Varity account
`varity_search_docs`	Search Varity documentation
`varity_cost_calculator`	Estimate costs for your project
`varity_install_deps`	Install project dependencies
`varity_dev_server`	Start a local development server
`varity_build`	Build your project
`varity_open_browser`	Open a URL in your browser
`varity_create_repo`	Create a GitHub repo and push your code
`varity_deploy`	Deploy to production
`varity_deploy_status`	Check deployment status
`varity_deploy_logs`	View build and runtime logs
`varity_migrate`	Migrate an existing app from Vercel to Varity

Set up the MCP server to get started.

Next steps

Quick Start - Deploy your first app
Installation - Full setup guide
MCP Server Setup - Connect your AI coding tool
Vercel Migration - Move an existing app to Varity