How-to: get the SSH connection line and runtime ports for a running pod

This is a worked task for reading the SSH address of a running pod. The reference page (rp doc ssh info) documents the individual flags; this page shows the call composed into a single workflow.

Goal: get the ssh connection line and the runtime ports for a pod that is currently running, so you can reach it over SSH.

Steps

Get the raw pod id. rp ssh info takes the id verbatim — it does not resolve pod names — so read it from rp pod list first:

$ rp pod list

The ID column (for example rp_abc123) is the value to pass in step 2.

Print the connection line. Pass that raw id to rp ssh info; it reads the pod over the REST v2 control plane and prints the ssh line:

$ rp ssh info rp_abc123 ssh root@57.1.2.3 -p 22981

The login user defaults to root (Runpod official images run as root).

Set the user for a non-root image. Images that start as a non-root user need --user to match, otherwise the printed line fails with a permission error:

$ rp ssh info rp_abc123 --user ubuntu ssh ubuntu@57.1.2.3 -p 22981

The --user value is not validated against the pod; pass the user the image actually starts as.

See the full pod record. --json prints the entire pod object the line was derived from — useful for inspecting every runtime port:

$ rp ssh info rp_abc123 --json

Notes

The connection line is built from the first runtime port whose type is ssh or tcp, in the order the API returns them. It does not guarantee an ssh-labelled port is preferred over a tcp one — whichever appears first in runtime.ports is used.
A stopped pod has no runtime and prints pod has no runtime (stopped?) rather than failing; start the pod and ask again.
A running pod that exposes no ssh or tcp port prints its available runtime ports instead of a connection line.
Registering a key with rp ssh-key add is what makes the address reachable; rp ssh info only reports it.
The call rides GET /v2/pods/{id} on the control plane (api.runpod.io/v2).