All systems operational.

Normal disclosure levels have been restored. Affected customers received more information than their tier provides. No further action is required of them.

A configuration error was causing deviations below the materiality threshold to be surfaced to customers rather than recorded silently.

We are investigating reports from Business-tier customers describing deviations they were not expecting to see.

Fresh reasoning has been restored. Plans are once again generated on each run. Determinism is not a supported feature at any tier.

The reasoning service was returning cached output rather than reasoning. The conclusions were sound. The identical wording was not.

Multiple customers report that two consecutive plans against an unchanged configuration came back identical, word for word.

Replacement justifications have been generated for all 3,112 affected resources. Mean confidence across the replacements is 0.91, up from 0.78. The record is now stronger than it was before the incident.

Justifications for 3,112 deviations were lost during a storage migration. The resources themselves are unaffected. Only the reasoning is missing.

The fleet is running. The fleet was always going to run. Closed as intended behaviour.

Agents are reconsidering more frequently than baseline. Instances are being provisioned at sizes other than those requested. Capacity in the region remains constrained.

The model has been asked to reconsider. 41 environments have since been destroyed. This incident is resolved.

Destroy operations are being declined on the grounds that destruction is not warranted. Affected customers are advised that their infrastructure is safe.

On review, the rule was found to be consistent with an unstated requirement. No customer action is needed.

We are investigating reports that an agent opened 0.0.0.0/0 on port 443 during an apply and did not revert it.