Skip to main content
Version: 0.9.20

Teardown

Teardown removes the AlphaAgent Studio infrastructure from your account. By design, your data is preserved unless you delete it explicitly — teardown removes the running services and supporting stacks, but retains your data stores so nothing is lost by accident. This page explains exactly what is deleted, what survives, and how to fully clean up if that is your goal.

Running teardown​

python3 studioctl.py teardown

studioctl asks for explicit confirmation, defaulting to no. On confirmation it deletes the deployment's stacks in reverse order (services first, foundation last).

[Screenshot: studioctl teardown confirmation prompt]

What is deleted vs. retained​

Retained (your data survives)​

These are intentionally kept so a teardown never destroys your data:

  • DynamoDB tables — agents, conversations, datasets, and all operational data.
  • S3 buckets — workspaces, uploaded documents, the web app, and other artifacts.
  • ElastiCache (Redis).
  • ECR repositories — your copy of the Studio image.
  • The activated license secret — so a later redeploy in the same account can resume the same license without re-activating.
  • Neo4j data — for self-hosted Neo4j, the encrypted EFS file system that holds the graph is retained (its Retain deletion policy survives teardown), so a redeploy can reattach to your existing graph; the stored credentials secret is retained too. For external Neo4j, your instance is entirely outside the installer's control.

Deleted​

  • ECS services, the Fargate cluster, the load balancer, and networking created by the stacks.
  • The Cognito user pool — users are federated shadows of your IdP and are recreated on the next deploy, so the pool is removed.
  • The one-time license bootstrap secret — it is spent after activation and should not linger.

Fully cleaning up (optional)​

If you want to remove everything, complete these manual steps after teardown — they are deliberately manual to prevent accidental data loss:

  1. Empty and delete the S3 buckets. Buckets must be emptied before they can be deleted.
  2. Delete the DynamoDB tables.
  3. Delete the ElastiCache (Redis) resources.
  4. Empty and delete the ECR repositories.
  5. Delete the retained license secret from Secrets Manager (only if you do not intend to redeploy with the same license).
  6. Deal with Neo4j. For self-hosted Neo4j, delete the retained EFS file system and the Neo4j credentials secret if you no longer need the graph (they are kept by default to protect your data). For external Neo4j, terminate your own instance — the installer cannot do this for you.
  7. Remove the small deploy helper bucket the installer created, if you no longer need it.

Redeploying into the same account​

Because data stores are retained, redeploying into the same account will collide with the retained resources unless you cleaned them up first. The preflight check in the deploy walkthrough is designed to catch exactly these collisions and tell you what already exists. To start completely fresh, do the full cleanup above before redeploying; to resume with your existing data, keep the retained resources in place.