ICE · Integrated Cloud Environment
Design cloud infrastructure visually. Ship it anywhere.
Drag a Postgres block onto the canvas. Connect it to a service. Wire a GitHub repo to a node. ICE compiles the canvas into real resources on AWS, Azure, or GCP. You get a cost estimate before deploying, live progress during, and metrics after. Claude can edit the canvas for you in plain English if you prefer.
What you get
One canvas covers the lifecycle of a cloud app.
Design, deploy, observe, predict cost, and switch providers without leaving the editor. Open source, scriptable, and yours to fork.
- Logs from running services stream into the canvas. Filter, search, follow.
- Live logs
- CPU, memory, request rate, latency. Sparklines next to the resource they describe.
- Per-block metrics
- Cloud Run, Lambda, App Service. Scale to zero is the default, not a feature you have to wire up.
- Scale to zero
- Per-canvas monthly projection before you click deploy. Drift alerts after.
- Cost before apply
Why ICE exists
Cloud infrastructure should be as visual as code is in an IDE.
ICE is built around a graph engine that does not know about specific clouds. Each cloud has its own deployer that turns the canvas into real resources, so you can swap the target without rewriting the design. Apache 2.0, no telemetry, no licence keys. Runs entirely on your machine if you want it to.
"I spent a decade watching teams pay the same complexity tax on cloud infrastructure. Different tools for design, deploy, observe. Different mental models. All hand-wired. ICE is what I wished existed: one canvas, one editor, deploy anywhere. It is open source now, so anyone can pick it up and ship."