Modules
The queue of failures and deviations a person must look at. Core.
An incident is anything a human must look at before the system can be trusted: a check-in that did not match its shift, a feed that missed its expected arrival, an import that rejected rows, a survey that overran its time limit. One word across every producer and every surface.
incidents: the module and kind that raised it, a severity, the record it cites, the node and person it concerns, a payload, and the lifecycle open → assigned → resolved → verified.
It is core because every solution module raises incidents; an opt-in ledger would let a producer write into a table no tenant could see. An in-database producer raises one through a function no client can call, so an incident cannot be forged.
Assigning, resolving and verifying each move the lifecycle in order, within the caller's territory, stamping who and when. The person who resolved an incident cannot verify it.
/incidents: the queue, filtered to the caller's territory.
incidents.
incidents.read and incidents.manage.
Never. Every project links it at provisioning.