Build a useful investigation brief
Build the count matrix
List unread item IDs, group membership, badge value, list total, filter, tab, device, read action, timestamp, and refresh state. Include item-level and grouped cases. Record a fresh view and a stale view.
Trace one transition
Open or mark one item read and compare badge, list, other tab, and second device after each refresh. Change one filter or group condition. Record whether the count changes correctly but later, or never matches the contract.
Write the count brief
State first divergence, definition, sync boundary, and competing explanations. Recommend one bounded transition or refresh experiment with pass criteria. Do not infer a server read bug from one stale browser surface. Compare a fresh read transition with a delayed refresh.
What to carry forward
The investigation is complete when count definition and update boundary explain representative drift, or the first unknown is explicit. Stop with one reconciliation experiment. Preserve grouped, filtered, and multi-surface cases so a fix does not fit only the simplest list.
Technical background: W3C guidance.
Keep the decision with the work.
Use a Work Item in Aglet to record the problem, the evidence you have, and the next decision. Add an owner and priority, then keep updates in the discussion so the next person can follow the reasoning.
Create an account See the product workflow