DynamoDB table without point-in-time recovery
Fast to write, impossible to restore — no PITR, no encryption choice.
Press Run. The stack loaded here fails in a way worth seeing.
▸What the engine finds2 findings
Not examples — DynamoDB table without point-in-time recovery was run through the real engine when this page was built, and this is what it returned.
Without PITR there is no way back from a bad write.INTENT
▸ 1 informational finding — no point-in-time recovery
No point-in-time recoveryAwsSolutions-DDB3Sessions
Without point-in-time recovery a bad write or a bad deploy is permanent. There is no window in which to go back.
Full rule text
The DynamoDB table does not have Point-in-time Recovery enabled. DynamoDB continuous backups represent an additional layer of insurance against accidental loss of data on top of on-demand backups. The DynamoDB service can back up the data with per-second granularity and restore it to any single second from the time PITR was enabled up to the prior 35 days. AwsSolutions-DDB3 guide →
Or start from a stack that fails interestingly
The single most common AI-generated CDK mistake: a bucket left open to the world.
SSH open to the entire internet — the classic copy-paste ingress rule.
Website hosting straight off a bucket — no CDN, no TLS, no access logging.
A database with storage encryption off and backups barely configured.
Action "*" on Resource "*" — the permission grant that ends incident reviews.
A REST API wired to Lambda, wide open, with no access logging.
Fast to write, impossible to restore — no PITR, no encryption choice.
A CDN that will happily serve your site unencrypted, with no logging.
A queue with no server-side encryption and no dead-letter queue.
Network traffic no one can reconstruct after the fact.
Six characters, no symbols, no MFA — defaults nobody revisited.
The one that passes. Encrypted, recoverable, with a dead-letter queue — what good looks like.