You can create different types of custom recovery steps to include in recovery plans.

Custom recovery steps are either command recovery steps or message prompt steps.

Command Recovery Steps

Command recovery steps contain either top-level commands or per-virtual machine commands.

Top-Level Commands
Run on the Site Recovery Manager Server. For example, you might use these commands to power on physical devices or to redirect network traffic. You cannot run top-level commands on VMware Site Recovery on VMware Cloud on AWS.
Per-Virtual Machine Commands
Site Recovery Manager associates per-virtual machine commands with newly recovered virtual machines during the recovery process. You can use these commands to complete configuration tasks after powering on a virtual machine. You can run the commands either before or after powering on a virtual machine. Commands that you configure to run after the virtual machine is powered on can run either on the Site Recovery Manager Server or in the newly recovered virtual machine. You cannot run commands on Site Recovery Manager Server on VMware Cloud on AWS. Commands that run on the newly recovered virtual machine are run in the context of the user account that VMware Tools uses on the recovered virtual machine. Depending on the function of the command that you write, you might need to change the user account that VMware Tools uses on the recovered virtual machine.

Message Prompt Recovery Steps

Present a message in the Site Recovery Manager user interface during the recovery. You can use this message to pause the recovery and provide information to the user running the recovery plan. For example, the message can instruct users to perform a manual recovery task or to verify steps. The only action users can take in direct response to a prompt is to dismiss the message, which allows the recovery to continue.

Execution of Commands and Prompt Steps

For vSphere Replication protection groups, the first command or prompt (or custom) step added between Create Writeable Storage Snapshot and the first non-empty VM priority group starts in parallel with the step Create Writeable Storage Snapshot to address restart failure scenarios.