New project

What are you deploying?

Pick a starting point and velixir takes it from there. Most people are live in about two minutes, and it is free to start.

Start from a template A working app in the language you use, built and deployed for you

Node.js A zero-dependency HTTP server on Node 22. Deploy → Python Flask behind gunicorn on Python 3.13. Deploy → Go net/http on Go 1.23, standard library only. Deploy → .NET A minimal API on .NET 10. Deploy → Deno Deno.serve with TypeScript, no build step. Deploy → PHP PHP 8.4 behind nginx and FPM. Deploy → Ruby Sinatra on Ruby 3.4. Deploy → Rust A std-only HTTP listener on Rust 1.86. Deploy → Status page A public status page for your services, with no dependencies. Deploy → Notes on Postgres A notes app wired to managed Postgres. Deploy → Directus A headless CMS on a database you own. Deploy → n8n Workflow automation you host yourself. Deploy → AI chat A private, streaming AI chat for Mistral, OpenAI or any compatible API. Deploy → Background jobs A job queue with a live dashboard, on managed Valkey. Deploy → Discord bot A Discord bot with slash commands and a status page that hands you its invite link. Deploy → Django + Postgres A Django guestbook with the admin, on managed Postgres. Deploy → FastAPI + Postgres A todo API with interactive docs, on managed Postgres. Deploy → Remote MCP server Your own MCP server for Claude Code, Cursor and VS Code, behind a secret token. Deploy → Next.js + Postgres A Next.js 15 guestbook on managed Postgres, with Server Actions. Deploy → Telegram bot A Telegram bot on webhooks that sleeps between messages, so it runs on the Free plan. Deploy → Umami Cookie-free, privacy-friendly web analytics, with every page view kept in your own managed Postgres. Deploy → Hermes Agent A sandbox for Nous Research's Hermes Agent: an AI agent you chat with on Telegram, on the model of your choice. Its memory, skills and files reset whenever the app restarts. Deploy → OpenClaw An AI sandbox: try the OpenClaw assistant on your own model and Telegram bot in minutes. It starts fresh on every restart. Deploy → Steel Browser A headless Chrome sandbox for AI agents: sessions over REST, Puppeteer and Playwright over one secure WebSocket. Resets on every restart. Deploy →

Every template is a public repository you can read first. Browse them with details

Deploy a public GitHub repo Paste a link and velixir builds it, with nothing to set up on GitHub

A link to a branch or a folder works too, for a repository with several apps in it. velixir works out what it is built with, in any of eleven languages. Next, you create a free account and it starts building.

Private repository? Here's how to deploy it

A private repository deploys two ways, and neither gives velixir access to GitHub:

1. From your machine, with the CLI
```
curl -fsSL https://velixir.net/install.sh | sh
velixir login
velixir deploy
```
Run the last one in your project folder. On Windows, install with dotnet tool install --global Velixir.Cli.
2. On every push, with GitHub Actions

Create the app, then copy the workflow from its Deploys tab into .github/workflows/velixir.yml and add a velixir API key as a repository secret named VELIXIR_API_KEY. Each push to your branch runs the deploy inside GitHub.

Create your account How the workflow works

Deploy your own code One command from your project folder, in any of eleven languages

1. Install the CLI
```
curl -fsSL https://velixir.net/install.sh | sh
```
On Windows, or if you already have .NET: dotnet tool install --global Velixir.Cli
2. Sign in
```
velixir login
```
Opens your browser. New here? You can create the account from that page.
3. Deploy
```
velixir deploy
```
Run it in your project folder. The first run sets the app up, then builds it and gives you the URL.

Prefer CI? A GitHub Actions workflow runs the same deploy on every push to your branch.

Create your account or skip this and let velixir login do it

Let your AI agent deploy it Your agent ships what it built and hands you a link to keep it

No account or card needed to start. Your agent deploys the app and gives you a link to claim it, which keeps it on your account.

Paste this into your agent

Claude Code, Codex, Cursor, Cline, or any agent that can run commands and read a web page.

Deploy this app to velixir. Follow the "Deploy from an AI agent" steps at https://velixir.net/llms.txt, leave out .env files and other secrets, and when it is live, give me the app URL and the claim link.

Or add the velixir MCP server

Gives your agent deploy and logs tools, so it can ship, read what went wrong and try again.

Claude Code

claude mcp add velixir -- npx -y @velixir/mcp

Cursor, Windsurf, Cline or Claude Desktop (MCP settings)

{ "mcpServers": { "velixir": { "command": "npx", "args": ["-y", "@velixir/mcp"] } } }

How agent deploys work

Add a database Managed Postgres or Valkey in the EU, ready for any app to use

Postgres Backed up, with a connection string for your app Create → Valkey A Redis-compatible cache, no server to run Create →

Not sure yet? Create an account and decide later. Free plan, no card.