Aws Stackset Lifecycle

Write-side retirement operations for a CloudFormation StackSet, run from the management / delegated-admin account. Two MUTATING methods, each a single locked execution that writes one result resource (deleteInstances polls its operation to a terminal state; deleteStackSet is a direct delete).

deleteInstances runs DeleteStackInstances for an explicit batch of deployment targets (OUs + accounts) and regions, with an explicit retainStacks flag — use it batch-by-batch so each member is only briefly affected. A safety guard refuses to operate on a whole OU/root unless confirmWholeTarget: true is set, so a fat-fingered fleet-wide delete is impossible. deleteStackSet runs DeleteStackSet once the stackset is empty.

The official @swamp/aws/cloudformation/stack-set (Cloud Control) type deletes every instance at once with no batching and no RetainStacks control, so it cannot express a staged, low-blast-radius retirement. This model fills that gap using the native CloudFormation StackSets API, as the mutating sibling to the read-only @jentz/aws-stackset-audit and @jentz/aws-stackset-drift-detect.

MUTATING: needs cloudformation:DeleteStackInstances / cloudformation:DeleteStackSet plus the stackset admin role — a *-readonly profile cannot run these, by design.