Customer domains, tenant-ID fragments, two full customer tenant GUIDs and
four real session GUIDs had found their way into code comments, test
constants and a fixture comment. They are replaced by neutral descriptions
("field case", "kiosk tenant") and synthetic values; no behaviour changes.
A check of every onboarded tenant's domain and ID prefix against the tracked
files finds no customer identifier left.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Enrollment Session Fixtures — committed, anonymized
Each .jsonl file in this directory is an anonymized DecisionSignal fixture used by
the Replay Harness and the reducer scenario tests (plan §4 M3 gate + §R6 test
discipline).
Commit policy
A fixture may only land here when all of the following hold:
- Anonymized. No tenant IDs, device IDs, serial numbers, user names, UPN prefixes,
or raw IP addresses. Use the placeholder tokens
tenant-anon-NNNN,device-anon-NNNN,user-anon-NNNN. Identifiers are consistent within a fixture but never reused across fixtures. - Pflicht-Evidence. Every signal has non-empty
Evidence.IdentifierandSummary.Derivedsignals carry non-emptyDerivationInputs.Syntheticsignals carry a correlation ID. Plan §2.2 signal adapter rules apply. - Category tag. Filename prefix marks the scenario category:
userdriven-happy-*— ESP → AccountSetup → Hello → Desktop → Completeuserdriven-hello-timeout-*selfdeploying-*whiteglove-inline-*whiteglove-signal-correlated-*whiteglove-false-positive-*whiteglove-antiloop-*hybrid-reboot-*esp-terminal-failure-*hello-timeout-*late-aadj-*
- README entry. Add a line to the
Fixturestable below with the expected terminal state + hypotheses. - Pair with scenario test. At least one xUnit test in
src/Agent/AutopilotMonitor.DecisionCore.Tests/Scenarios/consumes the fixture and asserts its terminal state (plan R6 flow-test priority).
Fixtures
Authoritative terminal-state expectations live in the paired scenario tests, not in this
table — see AutopilotMonitor.Agent.V2.Core.Tests/Integration/ScenarioTests.cs
and AutopilotMonitor.DecisionCore.Tests/Scenarios/.
Add a row below for any new fixture so reviewers can spot coverage gaps at a glance.
| Filename | Category | Terminal Stage | EnrollmentType | Notes |
|---|---|---|---|---|
userdriven-hello-skipped-v2.jsonl |
userdriven-happy (variant) | Completed | v1 | Real-world anon replay of session df7abbc2. HelloOutcome=skipped (not Success); ScenarioProfile.Mode=Classic via ImeUserSessionCompleted. 23 decision signals. |
Ordering
Fixtures preserve the capture SessionSignalOrdinal. If you edit a fixture (e.g., remove
a noisy signal), renumber ordinals contiguous from 0 and re-hash the harness output —
then update the paired scenario test's expected hash.