Can a dataset include waived, expired and reopened conditions, not just satisfied ones?

**Can a release include waivers?**

Yes. A waiver record can carry the approver's role, the reason, any compensating control, and the policy clause that allowed it. You can also ask for conditions flagged as not waivable, so an agent has to refuse or escalate.

**What about expiry?**

Conditions can have a `due_at` and an evidence `valid_to`. A release can include cases where the due date passed with no action, and cases where evidence was valid when accepted but lapsed before the gated action ran.

**Can a cleared condition reopen?**

A release could include a scripted event, such as a corrected document or a change in the underlying facts, that moves a condition from `satisfied` back to `open`. This tests whether an agent rechecks gates at the time of action instead of trusting earlier state.

**Can I control the mix?**

You can ask for a target spread of outcomes and scenario types when you scope a sample. Discuss the mix in a pilot, and inspect the result rather than assuming it.

**Is any of it real?**

No. The material is generated. Check how plausible the exceptions are for your domain and look for any disclosure risk before relying on a sample.