A preview of the whole product, on every pull request.
Open a PR and slipway brings up its own copy of the project: every service, a database of its own, and the URLs posted right on the PR. It goes away when the PR closes.
Ship your whole stack.
Push a repo and slipway builds it, runs it next to its database and puts it on a URL. Every pull request gets a live copy of the whole thing.
$5 free usage · no credit card required
GitHub · GitLab
Connect either one. Pushes deploy.
Postgres · Redis
Managed, in the same project as your app.
Every pull request
Gets a preview of the whole stack.
$8.76
Per vCPU a month, billed by the second.
32 MCP tools
So your coding agent can deploy too.
Everything your app needs,
in one project.
Your services, their databases, the jobs that run at night and the buckets that hold uploads all live together and find each other by name. Nothing to glue together.
Every push deploys.
Push to your branch and slipway builds it, rolls it out and tells you when it's live, with every step timed. If a deploy fails, the last good one keeps serving.
Services find each other by name.
Public services get HTTPS on the first deploy. Everything else stays private, and
reaches a sibling at http://api, even from another repo.
Postgres and Redis, next to your app.
Add a database and pick its connection URL, host or password from any service. High availability is a switch, and disks can grow on their own.
Logs, metrics and alerts you didn't have to set up.
Every service streams its logs and charts its CPU, memory and requests from the first deploy. When one runs hot, the alert says what happened and offers the fix.
Quiet services sleep, and stop costing.
Choose an idle window and a service scales to zero when nobody is using it. The next request wakes it, and metering stops while it sleeps.
Web apps, workers, cron and buckets.
Build from a Dockerfile, from plain source in most languages, or run any public image. Each piece gets the shape it needs.
One-click apps, already wired.
n8n, Metabase, Umami, Uptime Kuma and the rest deploy into your project with their secrets generated and their database filled in. The admin tools attach to a database you already run.
Compared How slipway compares.
How slipway compares.
Render, Railway and Coolify, row by row.
Render and Railway are hosted platforms, like slipway. Coolify is open source and runs on servers you manage. Each cell comes from the vendor's own docs or pricing page.
| What you get | slipway hosted | Render Full comparison | Railway Full comparison | Coolify open source, self-hosted |
|---|---|---|---|---|
| Preview for every pull request | Full stack, own databases | Full stack on Pro | Full stack PR environments | Per app or Compose stack |
| Databases | ||||
| Cron jobs and workers | Cron jobs and workers | Cron jobs and workers | Cron on any service | Scheduled tasks |
| Idle services sleep | Wakes on the next request | Free instances only | Wakes on traffic | Not documented |
| Private networking by name | By plain service name | Internal hostnames | Internal DNS names | Container names |
| Logs and metrics built in | Logs and metrics | Logs and metrics | Logs and metrics | Logs, metrics opt-in |
| MCP server for coding agents | 32 tools, first-party | Hosted MCP server | Hosted MCP server | Built in at /mcp |
| Where it runs | Our cloud, or your own cluster on Pro | Render's cloud | Railway's cloud | Servers you run |
| Pricing | Per second, Hobby $5/mo | Per instance, Pro $25/mo | Per second, Hobby $5/mo | Free, or Cloud $5/mo |
Compared October 2026 from each vendor's docs and pricing pages. "Not documented" means we couldn't find it there. Scroll sideways to compare
Where each one fits.
Render and Railway are simple to run and host the whole stack, with full-stack previews on Render's Pro plan and PR environments on Railway. Coolify runs the whole stack on servers you manage, with previews per app.
slipway gives you all three without giving one up. It runs on Kubernetes, so it can run on your own cluster too.
Two rates price everything.
Nothing while it sleeps.
A Postgres, a worker and a Rails app pay the same per vCPU and per GiB, metered by the second on what you configure. No instance sizes to outgrow.
$8.76
per vCPU / month
$4.38
per GiB of memory / month
Free
$0 forever
$5 of usage to start, once. No card.
Up to 1 vCPU · 1 GiB per service
Hobby Popular
$5 / mo
Comes with $10 of usage every month.
Up to 8 vCPU · 8 GiB per service
Pro
$20 / seat / mo
Comes with $20 of usage every month.
Up to 32 vCPU · 32 GiB per service
- July 23, 2026How to reach services behind a corporate VPN from a deployed container
Your app is deployed on a container platform, but the API it needs sits behind a corporate VPN. You can't install the VPN client in the container, and you don't need to. Run the VPN as a userspace sidecar and route through it over SOCKS5.
- July 16, 2026The NAT rule we couldn't see: Tailscale, etcd, and iptables-legacy
Tailscale's invisible iptables-legacy NAT rule, a duplicate Talos hostname, and the conntrack cache that kept a correct fix from applying. Notes from growing slipway's cluster past one machine, and why cluster traffic now runs on kernel WireGuard.
- June 15, 2026Introducing slipway
Every pull request becomes a live, isolated environment on real Kubernetes, in a couple of minutes. Here is what we built, and why.
Straight answers.
Something missing? The docs go deeper, and the chat in the corner reaches a person.
What can I run on slipway?
Anything with a Dockerfile, a Node, Python, Go, Ruby, Rust or Elixir app built without one, or any public image. Each one runs as a web service, a private service, a worker, a cron job or a static site, next to managed Postgres, Redis and S3-compatible buckets.
Do I need a config file?
No. Pick a repo in the console and slipway keeps the settings for you. If you would rather keep the stack in code, add a slipway.yaml and the repo becomes the source of truth.
How do previews work?
Turn them on for a project and every pull request gets its own copy of it, databases included, with the URLs posted on the PR. Previews sleep when nobody is looking and are removed when the PR closes.
How is it billed?
By the second, on the CPU and memory you give each service: $8.76 per vCPU and $4.38 per GiB a month. Disks, backups and buckets are billed on what they store. A sleeping service costs nothing for compute.
What do I get for free?
$5 of usage, once, with no card. Services sleep after five idle minutes and wake on the next request, and each one can use up to 1 vCPU and 1 GiB.
Can my coding agent use it?
Yes. slipway runs a first-party MCP server with 32 tools, so Claude Code, Cursor and other clients can deploy a branch, wait for it to go live, read its logs and run commands in it. Tokens are scoped to one workspace and one role.
Ship your first service
before this tab closes.
Connect GitHub or GitLab, pick a repo, and one click puts it on a real URL. Every push after that deploys itself.
$5 free usage · no credit card required