Ployz

Ployz turns any Linux server into a deployment platform. It builds your app, gives it an https address, runs your databases and workers next to it, and spins up an environment for every pull request. Start with one server and add more when you need them.

Features

• Git push to deploy. Every push to your main branch builds and ships.
• Preview environments. Every pull request gets its own copy of your app at its own https address, Copy the database and the preview gets its own, empty, with a setup command to seed it. Send the link to your client. When the pull request merges, production gets the code and the preview's setting changes.
• Your whole app on one canvas. Web, workers, databases, variables and logs in one dashboard.
• Review before you deploy. Dashboard edits are staged until you press Deploy.
• No Dockerfile needed. Ployz detects and builds Node, Python, Ruby, PHP, Go, Rust, Java, Elixir and more. It can also use your Dockerfile or run any Docker image.
• HTTPS by default. Every app gets an https address on its first deploy. Custom domains are supported.
• More servers when you need them. Add a server with one command. Servers share a private network, and a service can run up to 50 replicas across them. No Kubernetes.
• Built for coding agents. Anything the dashboard does, the ployz CLI does too, with JSON output. Anything that would delete a service, a volume or a domain waits for your approval.
• No lock-in. Your apps keep running on your servers if Ployz Cloud goes away. The engine and the dashboard are both open source.

Quick start

You need a Linux server you can reach as root over SSH.

1. Install the CLI:

   curl -fsSL https://ployz.sh | sh
   # or
   brew install getployz/ployz/ployz
   

2. Sign in to Ployz Cloud:

   ployz login
   

3. From your app's directory, point Ployz at your server:

   ployz up --server root@your-server
   

   Ployz installs itself on the server, builds your app and prints its https address.

To deploy on every push and get preview environments, connect your repository with ployz github connect or from the dashboard at ployz.dev.

Using a coding agent? Paste Set me up for Ployz: https://ployz.sh/agent.md into it. The agent installs the CLI and the Ployz skill and signs you in.

How it works

flowchart LR
  you["Dashboard / ployz CLI"] -- deploys --> a
  subgraph servers["Your servers (private network)"]
    a["Server 1<br>your services"] <--> b["Server 2<br>your services"]
  end

Each server runs a small Ployz daemon alongside Docker. The servers form a private network and keep each other up to date, so there is no central control server to run or lose. The dashboard and the CLI tell your servers what to run; once deployed, your apps don't depend on either.

Ployz is built on proven parts: Docker for containers, WireGuard for the private network and Caddy for https.

Pricing

Ployz is free on your own servers, custom domains included. Paid Ployz Cloud plans are coming soon. Your hosting provider bills you for the servers.

Self-hosting the dashboard

You can run the dashboard yourself instead of using Ployz Cloud. See dashboard/README.md.

Status

Ployz is young and changing quickly. Good to know before you move a production app:

• You look after the server's OS. Ployz doesn't install OS updates.
• No database backups yet. Keep your own dumps off the server, or keep your managed database and connect to it with a variable.
• No automatic failover. If a server dies, the services on it stop. Replicas on other servers keep running.
• No Docker Compose files and no alerts. Use an uptime monitor for alerts.

Planned: snapshots, backups and rollbacks, and moving a database to a bigger server without downtime.

How Ployz is built

Most of Ployz is written by agents. I don't hand-edit much code anymore. I design the loop the agents run in, and I'm responsible for what comes out of it.

We use pstack. Every change ships as a small PR, and the agent has to prove it works by actually running Ployz the way you would: deploying to real servers, adding a machine, checking that traffic moves. If it can't show me that, it doesn't merge.

Low-risk PRs merge on their own after the checks pass. Higher-risk ones, like the orchestrator, networking, or anything touching your data, get a human review first. I read everything else after it lands.

When an agent gets something wrong, we don't just fix that one spot. We turn it into a lint rule or a test, so it can't happen again.

Code review never caught every bug, and verification won't either. But every mistake makes the system stricter, and that's the part I trust.

Contributing

Bug reports and feature requests go in GitHub Issues.

The repository has two projects: core/ (the engine, CLI and daemon, in Rust) and dashboard/ (Ployz Cloud, in TypeScript). Each README explains how to build and test it.

License

The engine in core/ is Apache-2.0. The dashboard in dashboard/ is AGPL-3.0.