Quickstart — Spec 118 Golden Master Deep Drift v2

Spec reference: /Users/ahmeddarrazi/Documents/projects/TenantAtlas/specs/118-baseline-drift-engine/spec.md

Prereqs

Format (dirty only): vendor/bin/sail bin pint --dirty --format agent
Tests (focused): vendor/bin/sail artisan test --compact --filter=Baseline (adjust filter to match added tests)

In /admin, open a Baseline Profile and set capture mode to Full content.
Run Capture baseline (full content) and select a source tenant.
Open Monitoring → Operations → the capture OperationRun:
- Verify context.target_scope renders a “Target” (no “No target scope details…”).
- Verify context.baseline_capture.evidence_capture and context.baseline_capture.gaps exist.
- If capture was incomplete, verify context.baseline_capture.resume_token exists and UI offers Resume capture.
In /admin/t/{tenant}, open Baseline Compare and run Compare now (full content).
Open Monitoring → the compare OperationRun:
- Verify coverage proof is recorded (context.baseline_compare.coverage).
- Verify evidence capture stats and evidence gaps are present.
- Verify “Why no findings?” is explained via context.baseline_compare.reason_code when applicable.
Open Findings:
- Verify drift findings appear when policy settings differ.
- Verify findings include fidelity/provenance fields for baseline and current evidence.

To test resume behavior without waiting for real throttling:

Temporarily reduce the per-run capture budget in config (planned in Spec 118) so the run cannot process the full scope.
Re-run capture/compare and verify a resume token is recorded and Resume capture continues work without duplicating already-captured subjects.