Skip to main content
The rp command-line tool runs the same Raypacks engine that builds your apps on Nubo, but locally. Useful for previewing a build, debugging detection, or producing an OCI image without pushing. You don’t need rp to deploy on Nubo. Pushing to your tracked branch is enough. Use rp when you want to inspect what Raypacks is about to do, or build a one-off image for somewhere else.

Install

One line:
The script detects your OS and CPU, downloads the right rp binary, and drops it in /usr/local/bin (or ~/.local/bin if it can’t get write access). Linux x86_64 and macOS arm64 (Apple Silicon) are published today.

Customizing the install

The installer reads a few env vars:

Manual download

If you’d rather grab the binary directly:
Tagged releases are at https://dl.withnubo.com/raypacks/<version>/rp-<os>-<arch>.

Verify

Commands

rp detect

Print which pack matches a project.
Useful when you’re not sure which pack Raypacks will pick. The summary names the detected stack and the entrypoint it’ll build.

rp plan

Show the full build plan without building.
You get the same plan table Nubo prints at the start of every cloud build:
The decision line shows where each setting came from: nubo.toml, --entrypoint, or auto. Flags:
  • --port <n>: override the runtime port
  • --entrypoint <path>: override the build entrypoint (Go pack only). nubo.toml still wins if it’s set.

rp build

Build an OCI image and write it to a tar (for local docker load) or push it to a registry.
Common flags:
  • --push <reference>: push to a registry instead of writing a tar
  • --tag <reference>: override the local tag (default raypacks-app:latest)
  • --output <path>: tar output path (default image.tar)
  • --port <n>: runtime port
  • --entrypoint <path>: override build entrypoint
  • --verbose: stream build step output instead of only on failure
RAYPACKS_REGISTRY_USER and RAYPACKS_REGISTRY_PASS work as env-var equivalents of the registry flags.

Loading a built image into Docker

You’ll have the same image Nubo would have built and deployed.