I have an existing Railway service that was previously configured using railway.json and is now sourced from a prebuilt Docker image.

Under Settings → Deploy, the service still shows a legacy Custom Start Command with the message “The value is set in /railway.json”, and the field is not editable.

The service is now image-sourced, so a new image deployment does not fetch the current GitHub repository or its railway.json. The new image already defines its production environment and Docker CMD, but I need to replace the retained legacy Custom Start Command with:

node dist/index.js

Railway's API documentation shows that serviceInstanceUpdate supports a startCommand field.

Before changing production, I’m trying to confirm the supported behavior:

1. Is serviceInstanceUpdate with startCommand the correct way to supersede or clear a legacy /railway.json-derived Start Command on an image-sourced service?
2. Does changing startCommand through that API merely stage a configuration change, or does it immediately trigger/redeploy the service?
3. Is there a supported dashboard method to remove the old config-as-code-derived Start Command without reconnecting the GitHub repository?
4. Can the Start Command correction and Docker image update be staged together and applied in one deployment?

I specifically want to avoid an intermediate deployment between the Start Command correction and image update.