# Diagnosis worksheet

Capture: symptom; expected behavior; trigger; version/environment; baseline evidence; hypotheses and distinguishing observations; experiment; actual result; cause; correction; original reproduction after correction; regression results; remaining uncertainty.

Example investigation:

| Hypothesis | Distinguishing observation | Result |
| --- | --- | --- |
| Browser sends twice | Same command ID in both deliveries | Repeated transport is expected |
| Writer allocates twice | Two record IDs for the same operation key | Confirmed at writer boundary |
| UI duplicates one record | One file/ID, two displayed rows | Not observed in this experiment |

Create a bug TSK if implementation is not already assigned. Include the reproduction and acceptance criterion, not only the proposed fix. Use LOG/DOC for the evidence packet and RDT for a substantial unresolved investigation. Preserve failed output and timestamps.

When reproduction is unavailable, say exactly which boundary could not be exercised, why, and the next safe experiment. Do not replace a missing real-provider observation with a fixture success claim.