[lucas-koontz]

User story

As the engineer who runs a customer PR environment
I want the close reminder to list everything the env leaves behind, and a blocked deploy to say how to retry without losing the env
So that nothing keeps running by accident, and a retry never costs the env its database

Why this matters

When a customer PR closes, the reminder lists its files, Cowork workspaces and Postgres volume, but not its SAM stack. The stack can outlive the PR, so it keeps running and holding the env's data until someone notices. When a Deploys: link conflicts, the blocked-deploy comment tells the author to retry by toggling the deploy label. Removing that label deletes the env, so the suggested retry recreates it on an empty database.

Test plan

• The close reminder names the SAM stack: a unit test renders the reminder for a closed customer PR, and its new line names anton-services-pr-<slug>-<N> and ends with "if it still exists".
• The blocked-deploy comment says how to fix a conflicted link and to re-run the failed jobs: a test renders the comment with the action's own printf through bash and checks both steps.
• The comment no longer suggests toggling deploy, and warns that removing it deletes the env: same test.
• The Postgres line no longer asks the operator to delete the volume's AWS Backup recovery points. The backup vault is locked, so nobody can. test_reminder_asks_for_nothing_the_backup_vault_refuses pins it; 252 tests pass.