bump pointer: internal (hello wait timeout plan, D-301)

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

scripts/CustomerSetup/Grant-AutopilotMonitorAddOn.ps1

@@ -75,15 +75,17 @@
 .NOTES
     Requires:
       - Microsoft.Graph.Authentication PowerShell module (auto-installs if missing).
-      - Sign-in as Global Administrator OR Privileged Role Administrator OR Cloud Application Administrator.
+      - Sign-in as Global Administrator OR Privileged Role Administrator (with PIM, activate the
+        role first). Only these roles may assign Microsoft Graph application permissions;
+        Application Administrator and Cloud Application Administrator are not enough.
+      - The Autopilot Monitor app must be admin-consented in your tenant first
+        (its service principal must exist before grants can be added).
 
     Sign-in behaviour:
       In Azure Cloud Shell the script signs in via the ambient identity endpoint (a delegated
       token for the already-signed-in user) - no device-code prompt, which Conditional Access
       policies commonly block. Outside Cloud Shell it falls back to a normal interactive
       Microsoft Graph sign-in. -TenantId only applies to the interactive fallback.
-      - The Autopilot Monitor app must be admin-consented in your tenant first
-        (its service principal must exist before grants can be added).
 
     After granting:
       The Autopilot Monitor backend may cache an older token for up to ~1 hour. Use the
@@ -139,6 +141,36 @@ if ($PSCmdlet.ParameterSetName -eq 'ByFeatures') {
 # ---- Constants ----
 $MicrosoftGraphAppId = '00000003-0000-0000-c000-000000000000'
 
+# ---- Graph failure details ----
+# Invoke-MgGraphRequest puts only "Response status code does not indicate success: ..." into the
+# exception message. Graph's own error (code + message) is the response body in ErrorDetails,
+# behind the request line, the status line and the headers. Some headers carry JSON themselves
+# (x-ms-ags-diagnostic), so the body is taken from behind the blank line that ends the headers.
+function Get-GraphFailure {
+    param([System.Management.Automation.ErrorRecord]$ErrorRecord)
+
+    $status = 0
+    if ($ErrorRecord.Exception.PSObject.Properties['Response'] -and $ErrorRecord.Exception.Response) {
+        $status = [int]$ErrorRecord.Exception.Response.StatusCode
+    }
+
+    $text = $ErrorRecord.Exception.Message
+    if ($ErrorRecord.ErrorDetails -and $ErrorRecord.ErrorDetails.Message) {
+        $body = ($ErrorRecord.ErrorDetails.Message -split '\r?\n\r?\n', 2)[-1].Trim()
+        if ($body.StartsWith('{')) {
+            try {
+                $graphError = ($body | ConvertFrom-Json).error
+                if ($graphError.code) { $text = "$($graphError.code): $($graphError.message)" }
+            }
+            catch {
+                # Not a Graph JSON body - keep the exception message.
+            }
+        }
+    }
+
+    [pscustomobject]@{ Status = $status; Text = $text }
+}
+
 # ---- Pre-flight: module ----
 if (-not (Get-Module -ListAvailable -Name Microsoft.Graph.Authentication)) {
     Write-Host "[Setup] Microsoft.Graph.Authentication not installed. Installing for current user..." -ForegroundColor Yellow
@@ -188,7 +220,7 @@ Write-Host "[Auth] Granted scopes: $($ctx.Scopes -join ', ')"
 # carries a fixed scope set (incl. Directory.AccessAsUser.All) instead of the requested scopes,
 # and its effective rights follow the signed-in user's directory role.
 if (-not $identityConnected -and -not ($ctx.Scopes -contains 'AppRoleAssignment.ReadWrite.All')) {
-    Write-Warning "Scope 'AppRoleAssignment.ReadWrite.All' was NOT granted. The signed-in user likely lacks the required role (Global Admin / Privileged Role Admin / Cloud App Admin)."
+    Write-Warning "Scope 'AppRoleAssignment.ReadWrite.All' was NOT granted. The signed-in user likely lacks the required role (Global Administrator or Privileged Role Administrator)."
     if (-not $VerifyOnly) { return }
 }
 
@@ -262,6 +294,7 @@ $action = if ($Revoke) { 'Revoke' } else { 'Grant' }
 Write-Host "[4/4] $action mode - processing $($Permissions.Count) permission(s)..." -ForegroundColor Cyan
 
 $results = @()
+$graphForbidden = $false
 foreach ($permName in $Permissions) {
     $appRole = $graphSp.appRoles | Where-Object {
         $_.value -eq $permName -and $_.allowedMemberTypes -contains 'Application'
@@ -288,7 +321,9 @@ foreach ($permName in $Permissions) {
                 $results += [pscustomobject]@{ Permission = $permName; Status = 'Revoked' }
             }
             catch {
-                Write-Warning "  [FAIL]  '$permName' revoke failed: $($_.Exception.Message)"
+                $failure = Get-GraphFailure $_
+                if ($failure.Status -eq 403) { $graphForbidden = $true }
+                Write-Warning "  [FAIL]  '$permName' revoke failed: $($failure.Text)"
                 $results += [pscustomobject]@{ Permission = $permName; Status = 'Failed' }
             }
         }
@@ -315,7 +350,9 @@ foreach ($permName in $Permissions) {
             $results += [pscustomobject]@{ Permission = $permName; Status = 'Granted' }
         }
         catch {
-            Write-Warning "  [FAIL]  '$permName' grant failed: $($_.Exception.Message)"
+            $failure = Get-GraphFailure $_
+            if ($failure.Status -eq 403) { $graphForbidden = $true }
+            Write-Warning "  [FAIL]  '$permName' grant failed: $($failure.Text)"
             $results += [pscustomobject]@{ Permission = $permName; Status = 'Failed' }
         }
     }
@@ -324,6 +361,9 @@ foreach ($permName in $Permissions) {
 Write-Host ""
 Write-Host "=========== Summary ===========" -ForegroundColor Cyan
 $results | Format-Table -AutoSize | Out-Host
+if ($graphForbidden) {
+    Write-Warning "Microsoft Graph refused the change (403): the signed-in account may not manage Microsoft Graph application permissions. Sign in as Global Administrator or Privileged Role Administrator (with PIM, activate the role first) and run the script again."
+}
 Write-Host ""
 Write-Host "Note: token-cached backends may need up to ~1h to see the new permission set." -ForegroundColor Yellow
 Write-Host "      Use 'Refresh permission status' in the admin UI to apply changes immediately." -ForegroundColor Yellow

src/Agent/AutopilotMonitor.Agent.V2.Core.Tests/Monitoring/HelloTrackerCompletionBudgetTests.cs

@@ -0,0 +1,71 @@
+#nullable enable
+using System;
+using System.Collections.Generic;
+using System.Linq;
+using AutopilotMonitor.Agent.V2.Core.Logging;
+using AutopilotMonitor.Agent.V2.Core.Monitoring.Enrollment.SystemSignals;
+using AutopilotMonitor.Agent.V2.Core.Orchestration;
+using AutopilotMonitor.Agent.V2.Core.Tests.Harness;
+using AutopilotMonitor.Agent.V2.Core.Tests.Orchestration;
+using AutopilotMonitor.DecisionCore.Engine;
+using AutopilotMonitor.DecisionCore.Signals;
+using AutopilotMonitor.Shared;
+using Xunit;
+
+namespace AutopilotMonitor.Agent.V2.Core.Tests.Monitoring
+{
+    /// <summary>
+    /// The completion timer a started Hello wizard runs against is the tenant's Hello wait
+    /// (<c>HelloWaitTimeoutSeconds</c>), resolved to 300..3600 s — the same budget the engine's
+    /// hello_safety window spends from the ESP exit, so a tenant allowing an hour is not cut off
+    /// by a fixed 5-min tracker timer while the user is still inside the wizard.
+    /// </summary>
+    public sealed class HelloTrackerCompletionBudgetTests
+    {
+        private static readonly DateTime Fixed = new DateTime(2026, 9, 29, 8, 0, 0, DateTimeKind.Utc);
+
+        [Theory]
+        [InlineData(30, 300)]     // the old default, still stored for many tenants → the built-in window
+        [InlineData(0, 300)]
+        [InlineData(300, 300)]
+        [InlineData(1800, 1800)]
+        [InlineData(3600, 3600)]
+        [InlineData(86400, 3600)] // clamped to the ceiling
+        public void Completion_budget_is_the_resolved_tenant_hello_wait(int configured, int expected)
+        {
+            using var tmp = new TempDirectory();
+            using var tracker = NewTracker(tmp, new FakeSignalIngressSink(), configured);
+
+            Assert.Equal(expected, tracker.CompletionTimeoutSecondsForTest);
+        }
+
+        [Fact]
+        public void Completion_timeout_event_reports_the_tenant_budget()
+        {
+            using var tmp = new TempDirectory();
+            var ingress = new FakeSignalIngressSink();
+            using var tracker = NewTracker(tmp, ingress, helloWaitTimeoutSeconds: 2400);
+
+            tracker.NotifyHelloWizardStarted();
+            Assert.True(tracker.IsCompletionTimerActiveForTest);
+            tracker.TriggerCompletionTimeoutForTest();
+
+            var timeout = ingress.Posted.Single(p =>
+                p.Kind == DecisionSignalKind.InformationalEvent
+                && p.Payload != null
+                && p.Payload.TryGetValue(SignalPayloadKeys.EventType, out var et)
+                && et == Constants.EventTypes.HelloCompletionTimeout);
+            var data = Assert.IsAssignableFrom<IDictionary<string, object>>(timeout.TypedPayload);
+            Assert.Equal(2400, data["timeoutSeconds"]);
+            Assert.Equal("timeout", tracker.HelloOutcome);
+        }
+
+        private static HelloTracker NewTracker(TempDirectory tmp, FakeSignalIngressSink ingress, int helloWaitTimeoutSeconds) =>
+            new HelloTracker(
+                sessionId: "S1",
+                tenantId: "T1",
+                post: new InformationalEventPost(ingress, new VirtualClock(Fixed)),
+                logger: new AgentLogger(tmp.Path, AgentLogLevel.Info),
+                helloWaitTimeoutSeconds: helloWaitTimeoutSeconds);
+    }
+}

src/Agent/AutopilotMonitor.Agent.V2.Core.Tests/Monitoring/SystemSignals/ProvisioningStatusTrackerAccountSetupActivityTests.cs

@@ -83,6 +83,7 @@ namespace AutopilotMonitor.Agent.V2.Core.Tests.Monitoring.SystemSignals
             using var f = new Fixture();
             Assert.False(f.Tracker.HasAccountSetupActivity);
             Assert.False(f.Tracker.HasAccountSetupProgress);
+            Assert.Null(f.Tracker.AccountSetupProgressIfRead);
         }
 
         [Fact]
@@ -97,6 +98,8 @@ namespace AutopilotMonitor.Agent.V2.Core.Tests.Monitoring.SystemSignals
 
             Assert.True(f.Tracker.HasAccountSetupActivity);
             Assert.False(f.Tracker.HasAccountSetupProgress);
+            // Read and untouched — the only state the WhiteGlove sealing gate treats as "false".
+            Assert.False(f.Tracker.AccountSetupProgressIfRead);
         }
 
         [Fact]
@@ -108,6 +111,7 @@ namespace AutopilotMonitor.Agent.V2.Core.Tests.Monitoring.SystemSignals
 
             Assert.True(f.Tracker.HasAccountSetupActivity);
             Assert.True(f.Tracker.HasAccountSetupProgress);
+            Assert.True(f.Tracker.AccountSetupProgressIfRead);
         }
 
         [Theory]
@@ -138,5 +142,18 @@ namespace AutopilotMonitor.Agent.V2.Core.Tests.Monitoring.SystemSignals
 
             Assert.False(f.Tracker.HasAccountSetupProgress);
         }
+
+        [Fact]
+        public void OnlyDeviceSetupRead_AccountSetupProgressIsUnknown()
+        {
+            // Adversarial for the WhiteGlove sealing gate: the tracker holds DeviceSetup data but
+            // has not read the AccountSetup category — that must be "unknown", never "untouched".
+            using var f = new Fixture();
+            f.Tracker.ProcessCategoryStatusForTest("DeviceSetupCategory.Status",
+                "{\"categoryState\":\"inProgress\",\"DeviceSetup.AppsSubcategory\":{\"subcategoryState\":\"inProgress\"}}");
+
+            Assert.Null(f.Tracker.AccountSetupProgressIfRead);
+            Assert.False(f.Tracker.HasAccountSetupProgress);
+        }
     }
 }

src/Agent/AutopilotMonitor.Agent.V2.Core.Tests/Monitoring/SystemSignals/StallProbeHelloWaitGuardTests.cs

@@ -0,0 +1,123 @@
+#nullable enable
+using System;
+using System.Linq;
+using AutopilotMonitor.Agent.V2.Core.Logging;
+using AutopilotMonitor.Agent.V2.Core.Monitoring.Enrollment.SystemSignals;
+using AutopilotMonitor.Agent.V2.Core.Orchestration;
+using AutopilotMonitor.Agent.V2.Core.Tests.Harness;
+using AutopilotMonitor.Agent.V2.Core.Tests.Orchestration;
+using AutopilotMonitor.DecisionCore.Engine;
+using AutopilotMonitor.DecisionCore.Signals;
+using AutopilotMonitor.Shared;
+using Xunit;
+
+namespace AutopilotMonitor.Agent.V2.Core.Tests.Monitoring.SystemSignals
+{
+    /// <summary>
+    /// A tenant-extended Hello wait runs up to 60 min — the default session_stalled threshold. While
+    /// it is pending the session waits on a bounded timer that resolves it either way, so the stall
+    /// probe holds the report back instead of flagging the session Stalled; once the wait is over
+    /// and the session is still idle the report fires, and real activity drops it. The collector
+    /// runs with no scan sources, so only the session_stalled decision is exercised.
+    /// </summary>
+    public sealed class StallProbeHelloWaitGuardTests
+    {
+        private static readonly DateTime Fixed = new DateTime(2026, 9, 29, 8, 0, 0, DateTimeKind.Utc);
+        private const double IdleAtStallThreshold = 60;
+
+        [Fact]
+        public void Session_stalled_is_held_back_while_the_hello_wait_is_pending()
+        {
+            using var tmp = new TempDirectory();
+            var ingress = new FakeSignalIngressSink();
+            var collector = NewCollector(tmp, ingress, helloWaitPending: () => true);
+
+            collector.CheckAndRunProbes(IdleAtStallThreshold);
+            collector.CheckAndRunProbes(IdleAtStallThreshold + 1);
+
+            Assert.Equal(0, StalledCount(ingress));
+        }
+
+        [Fact]
+        public void Held_back_report_fires_once_the_hello_wait_is_over_and_the_session_is_still_idle()
+        {
+            using var tmp = new TempDirectory();
+            var ingress = new FakeSignalIngressSink();
+            var pending = true;
+            var collector = NewCollector(tmp, ingress, helloWaitPending: () => pending);
+
+            collector.CheckAndRunProbes(IdleAtStallThreshold);
+            Assert.Equal(0, StalledCount(ingress));
+
+            // hello_safety resolved (a deadline fire is no activity, so no ResetProbes): the next
+            // tick emits the held-back report — once.
+            pending = false;
+            collector.CheckAndRunProbes(IdleAtStallThreshold + 1);
+            collector.CheckAndRunProbes(IdleAtStallThreshold + 2);
+
+            Assert.Equal(1, StalledCount(ingress));
+        }
+
+        [Fact]
+        public void Real_activity_drops_the_held_back_report()
+        {
+            using var tmp = new TempDirectory();
+            var ingress = new FakeSignalIngressSink();
+            var pending = true;
+            var collector = NewCollector(tmp, ingress, helloWaitPending: () => pending);
+
+            collector.CheckAndRunProbes(IdleAtStallThreshold);
+            pending = false;
+            collector.ResetProbes();
+            collector.CheckAndRunProbes(1);
+            Assert.Equal(0, StalledCount(ingress));
+
+            // A new idle window reaches the threshold again — the fire-once report is still unused.
+            collector.CheckAndRunProbes(IdleAtStallThreshold);
+            Assert.Equal(1, StalledCount(ingress));
+        }
+
+        [Fact]
+        public void Without_a_hello_wait_probe_session_stalled_fires_as_before()
+        {
+            using var tmp = new TempDirectory();
+            var ingress = new FakeSignalIngressSink();
+            var collector = NewCollector(tmp, ingress, helloWaitPending: null);
+
+            collector.CheckAndRunProbes(IdleAtStallThreshold);
+
+            Assert.Equal(1, StalledCount(ingress));
+        }
+
+        [Fact]
+        public void A_throwing_hello_wait_probe_never_costs_the_stall_report()
+        {
+            using var tmp = new TempDirectory();
+            var ingress = new FakeSignalIngressSink();
+            var collector = NewCollector(tmp, ingress, helloWaitPending: () => throw new InvalidOperationException("probe down"));
+
+            collector.CheckAndRunProbes(IdleAtStallThreshold);
+
+            Assert.Equal(1, StalledCount(ingress));
+        }
+
+        private static StallProbeCollector NewCollector(TempDirectory tmp, FakeSignalIngressSink ingress, Func<bool>? helloWaitPending) =>
+            new StallProbeCollector(
+                sessionId: "S1",
+                tenantId: "T1",
+                post: new InformationalEventPost(ingress, new VirtualClock(Fixed)),
+                logger: new AgentLogger(tmp.Path, AgentLogLevel.Info),
+                thresholdsMinutes: new[] { 60 },
+                traceIndices: Array.Empty<int>(),
+                sources: Array.Empty<string>(),
+                sessionStalledAfterProbeIndex: 1,
+                helloWaitPending: helloWaitPending);
+
+        private static int StalledCount(FakeSignalIngressSink ingress) =>
+            ingress.Posted.Count(p =>
+                p.Kind == DecisionSignalKind.InformationalEvent
+                && p.Payload != null
+                && p.Payload.TryGetValue(SignalPayloadKeys.EventType, out var et)
+                && et == Constants.EventTypes.SessionStalled);
+    }
+}

src/Agent/AutopilotMonitor.Agent.V2.Core.Tests/Monitoring/Transport/EmergencyReporterTests.cs

@@ -20,6 +20,7 @@ namespace AutopilotMonitor.Agent.V2.Core.Tests.Monitoring.Transport
         private sealed class FakeApiClient : BackendApiClient
         {
             public readonly List<TimeSpan?> Timeouts = new List<TimeSpan?>();
+            public readonly List<AgentErrorReport> Reports = new List<AgentErrorReport>();
             public readonly Queue<bool> Results = new Queue<bool>();
 
             public FakeApiClient() : base(
@@ -40,6 +41,7 @@ namespace AutopilotMonitor.Agent.V2.Core.Tests.Monitoring.Transport
             public override Task<bool> ReportAgentErrorAsync(AgentErrorReport report, TimeSpan? timeout = null)
             {
                 Timeouts.Add(timeout);
+                Reports.Add(report);
                 return Task.FromResult(Results.Count > 0 && Results.Dequeue());
             }
         }
@@ -139,5 +141,37 @@ namespace AutopilotMonitor.Agent.V2.Core.Tests.Monitoring.Transport
 
             Assert.Equal(2, client.Calls);
         }
+
+        [Fact]
+        public async Task Session_age_is_carried_in_the_report_and_absent_by_default()
+        {
+            // One reporter per report: a delivered report starts the 10-minute cooldown.
+            var client = new FakeApiClient();
+            client.Results.Enqueue(true);
+            client.Results.Enqueue(true);
+
+            await Reporter(client).TrySendAsync(AgentErrorType.SessionAgeEmergencyBreak, "m", sessionAgeHours: 478.2);
+            await Reporter(client).TrySendAsync(AgentErrorType.ConfigFetchFailed, "m");
+
+            Assert.Equal(2, client.Reports.Count);
+            Assert.Equal(478.2, client.Reports[0].SessionAgeHours);
+            Assert.Null(client.Reports[1].SessionAgeHours);
+        }
+
+        [Fact]
+        public async Task Prior_registration_failure_is_carried_in_the_report_and_absent_by_default()
+        {
+            var client = new FakeApiClient();
+            client.Results.Enqueue(true);
+            client.Results.Enqueue(true);
+            var record = new RegistrationFailureSummary { FailedRuns = 2, Outcome = "Failed", NetworkLinkUpAtEnd = false };
+
+            await Reporter(client).TrySendAsync(AgentErrorType.SessionAgeEmergencyBreak, "m", priorRegistrationFailure: record);
+            await Reporter(client).TrySendAsync(AgentErrorType.ConfigFetchFailed, "m");
+
+            Assert.Equal(2, client.Reports.Count);
+            Assert.Same(record, client.Reports[0].PriorRegistrationFailure);
+            Assert.Null(client.Reports[1].PriorRegistrationFailure);
+        }
     }
 }

src/Agent/AutopilotMonitor.Agent.V2.Core.Tests/Orchestration/RecoveryCoordinatorTests.cs

@@ -162,6 +162,28 @@ namespace AutopilotMonitor.Agent.V2.Core.Tests.Orchestration
             Assert.Equal(SessionStage.AwaitingHello, result.InitialState.Stage);
         }
 
+        [Fact]
+        public void Clean_snapshot_of_a_rotated_session_adopts_the_current_session_id()
+        {
+            // SESSION-OWNER-BINDING: the backend refused the persisted session id at registration,
+            // the agent rotated, and the snapshot on disk still names the old session. The
+            // enrollment state carries over; the session id must be the current one.
+            using var rig = new Rig();
+            var saved = DecisionState.CreateInitial("S-REFUSED", TenantId)
+                .ToBuilder()
+                .WithStage(SessionStage.EspAccountSetup)
+                .WithStepIndex(7)
+                .WithLastAppliedSignalOrdinal(6)
+                .Build();
+            SeedSnapshot(rig, saved);
+
+            var result = rig.Recover();
+
+            Assert.Equal(SessionId, result.InitialState.SessionId);
+            Assert.Equal(SessionStage.EspAccountSetup, result.InitialState.Stage);
+            Assert.Equal(7, result.InitialState.StepIndex);
+        }
+
         [Fact]
         public void Snapshot_plus_pending_tail_replays_tail_onto_snapshot()
         {

src/Agent/AutopilotMonitor.Agent.V2.Core.Tests/Program/BackendSessionRegistrationTests.cs

@@ -118,5 +118,88 @@ namespace AutopilotMonitor.Agent.V2.Core.Tests.Program
             Assert.Throws<ArgumentNullException>(
                 () => BackendSessionRegistration.Register(config, auth, http, "1.0", consoleMode: false, logger: null));
         }
+
+        // ============================================================ failed-registration record
+
+        [Fact]
+        public void RecordRegistrationRun_records_a_failed_run_with_its_window_and_bounded_error()
+        {
+            using var tmp = new TempDirectory();
+            var persistence = new SessionIdPersistence(tmp.Path);
+            var started = new DateTime(2026, 8, 20, 9, 0, 0, DateTimeKind.Utc);
+
+            BackendSessionRegistration.RecordRegistrationRun(
+                persistence, SessionRegistrationResult.Failed(new string('e', 1000)),
+                started, started.AddSeconds(1204.46), linkUpAtStart: false, linkUpAtEnd: true,
+                configFetchOutcome: "UsedDefaults", logger: NewLogger(tmp.Path));
+
+            var record = persistence.LoadRegistrationFailure();
+            Assert.NotNull(record);
+            Assert.Equal(1, record!.FailedRuns);
+            Assert.Equal("Failed", record.Outcome);
+            Assert.Equal(BackendSessionRegistration.MaxRecordedErrorLength, record.LastError!.Length);
+            Assert.False(record.NetworkLinkUpAtStart);
+            Assert.True(record.NetworkLinkUpAtEnd);
+            Assert.Equal(1204.5, record.AttemptWindowSeconds);
+            Assert.Equal("UsedDefaults", record.ConfigFetchOutcome);
+            Assert.Equal(started.AddSeconds(1204.46), record.LastFailedAtUtc);
+        }
+
+        [Fact]
+        public void RecordRegistrationRun_records_auth_failures_and_clears_on_success()
+        {
+            using var tmp = new TempDirectory();
+            var persistence = new SessionIdPersistence(tmp.Path);
+            var at = new DateTime(2026, 8, 20, 9, 0, 0, DateTimeKind.Utc);
+
+            BackendSessionRegistration.RecordRegistrationRun(
+                persistence, SessionRegistrationResult.AuthFailed(401, "unauthorized"),
+                at, at.AddSeconds(3), true, true, "Succeeded", NewLogger(tmp.Path));
+            Assert.Equal("AuthFailed", persistence.LoadRegistrationFailure()!.Outcome);
+
+            BackendSessionRegistration.RecordRegistrationRun(
+                persistence, SessionRegistrationResult.Succeeded(new RegisterSessionResponse { Success = true }),
+                at, at.AddSeconds(1), true, true, "Succeeded", NewLogger(tmp.Path));
+            Assert.Null(persistence.LoadRegistrationFailure());
+        }
+
+        [Fact]
+        public void RecordRegistrationRun_without_persistence_is_a_no_op()
+        {
+            using var tmp = new TempDirectory();
+            BackendSessionRegistration.RecordRegistrationRun(
+                null, SessionRegistrationResult.Failed("x"), DateTime.UtcNow, DateTime.UtcNow,
+                null, null, null, NewLogger(tmp.Path));
+        }
+
+        // ============================================================ emergency-break message
+
+        [Fact]
+        public void Break_message_names_failed_registrations_in_invariant_culture()
+        {
+            var previous = System.Threading.Thread.CurrentThread.CurrentCulture;
+            try
+            {
+                System.Threading.Thread.CurrentThread.CurrentCulture = new System.Globalization.CultureInfo("de-DE");
+                var record = new RegistrationFailureSummary { FailedRuns = 3, Outcome = "Failed", NetworkLinkUpAtEnd = false };
+
+                var message = EmergencyBreakReporter.BuildMessage(478.2, 48, record);
+
+                Assert.Contains("session age 478.2h (cap 48h)", message);
+                Assert.EndsWith(" 3 earlier start(s) failed to register (last: Failed, network link at end: down).", message);
+            }
+            finally
+            {
+                System.Threading.Thread.CurrentThread.CurrentCulture = previous;
+            }
+        }
+
+        [Fact]
+        public void Break_message_without_record_keeps_the_historical_text()
+        {
+            Assert.Equal(
+                "Agent absolute session-age emergency break fired at session age 50.0h (cap 48h) — cleaning up and exiting.",
+                EmergencyBreakReporter.BuildMessage(50.0, 48, null));
+        }
     }
 }

src/Agent/AutopilotMonitor.Agent.V2.Core.Tests/Program/PostEnrollmentFactsObservedSignalTests.cs

@@ -4,6 +4,7 @@ using AutopilotMonitor.Agent.V2.Core.Logging;
 using AutopilotMonitor.Agent.V2.Core.Orchestration;
 using AutopilotMonitor.Agent.V2.Core.Tests.Harness;
 using AutopilotMonitor.Agent.V2.Core.Tests.Orchestration;
+using AutopilotMonitor.DecisionCore.Engine;
 using AutopilotMonitor.DecisionCore.Signals;
 using AutopilotMonitor.Shared.Models;
 using Xunit;
@@ -29,7 +30,7 @@ namespace AutopilotMonitor.Agent.V2.Core.Tests.Program
             var sink = new FakeSignalIngressSink();
             var logger = NewLogger(tmp.Path);
 
-            AutopilotMonitor.Agent.V2.Runtime.LifecycleEmitters.PostEnrollmentFactsObserved(sink, logger);
+            AutopilotMonitor.Agent.V2.Runtime.LifecycleEmitters.PostEnrollmentFactsObserved(sink, HelloWaitTimeout.DefaultSeconds, logger);
 
             Assert.Single(sink.Posted);
             var posted = sink.Posted[0];
@@ -47,7 +48,7 @@ namespace AutopilotMonitor.Agent.V2.Core.Tests.Program
             var sink = new FakeSignalIngressSink();
             var logger = NewLogger(tmp.Path);
 
-            AutopilotMonitor.Agent.V2.Runtime.LifecycleEmitters.PostEnrollmentFactsObserved(sink, logger);
+            AutopilotMonitor.Agent.V2.Runtime.LifecycleEmitters.PostEnrollmentFactsObserved(sink, HelloWaitTimeout.DefaultSeconds, logger);
 
             var payload = sink.Posted[0].Payload;
             Assert.NotNull(payload);
@@ -73,6 +74,34 @@ namespace AutopilotMonitor.Agent.V2.Core.Tests.Program
             Assert.True(cloudPc == "true" || cloudPc == "false");
         }
 
+        [Theory]
+        [InlineData(300)] // the default — the payload stays exactly what it was before the key existed
+        [InlineData(30)]  // the old default, still stored for many tenants
+        [InlineData(0)]
+        public void Hello_wait_not_extending_the_built_in_window_is_not_stamped(int helloWaitTimeoutSeconds)
+        {
+            using var tmp = new TempDirectory();
+            var sink = new FakeSignalIngressSink();
+
+            AutopilotMonitor.Agent.V2.Runtime.LifecycleEmitters.PostEnrollmentFactsObserved(sink, helloWaitTimeoutSeconds, NewLogger(tmp.Path));
+
+            Assert.DoesNotContain(SignalPayloadKeys.HelloWaitTimeoutSeconds, (IDictionary<string, string>)sink.Posted[0].Payload!);
+        }
+
+        [Theory]
+        [InlineData(1800, "1800")]
+        [InlineData(3600, "3600")]
+        [InlineData(86400, "3600")] // the effective (clamped) value is what the engine should arm
+        public void Extended_hello_wait_is_stamped_as_its_effective_value(int helloWaitTimeoutSeconds, string expected)
+        {
+            using var tmp = new TempDirectory();
+            var sink = new FakeSignalIngressSink();
+
+            AutopilotMonitor.Agent.V2.Runtime.LifecycleEmitters.PostEnrollmentFactsObserved(sink, helloWaitTimeoutSeconds, NewLogger(tmp.Path));
+
+            Assert.Equal(expected, sink.Posted[0].Payload![SignalPayloadKeys.HelloWaitTimeoutSeconds]);
+        }
+
         [Fact]
         public void Swallows_sink_exceptions()
         {
@@ -82,7 +111,7 @@ namespace AutopilotMonitor.Agent.V2.Core.Tests.Program
             var logger = NewLogger(tmp.Path);
 
             var ex = Record.Exception(() =>
-                AutopilotMonitor.Agent.V2.Runtime.LifecycleEmitters.PostEnrollmentFactsObserved(sink, logger));
+                AutopilotMonitor.Agent.V2.Runtime.LifecycleEmitters.PostEnrollmentFactsObserved(sink, HelloWaitTimeout.DefaultSeconds, logger));
 
             Assert.Null(ex);
         }

src/Agent/AutopilotMonitor.Agent.V2.Core.Tests/Program/ProgramGuardsTests.cs

@@ -149,6 +149,7 @@ namespace AutopilotMonitor.Agent.V2.Core.Tests.Program
             persistence.SaveSessionCreatedAt(DateTime.UtcNow.AddHours(-100));
 
             var stateDir = Path.Combine(tmp.Path, "State");
+            double? reportedAgeHours = null;
 
             var tripped = AutopilotMonitor.Agent.V2.Program.CheckSessionAgeEmergencyBreak(
                 dataDirectory: tmp.Path,
@@ -157,9 +158,13 @@ namespace AutopilotMonitor.Agent.V2.Core.Tests.Program
                 selfDestructOnComplete: false,
                 cleanupServiceFactory: null,
                 logger: logger,
-                consoleMode: false);
+                consoleMode: false,
+                onBreakFired: ageHours => reportedAgeHours = ageHours);
 
             Assert.True(tripped);
+            // The measured age reaches the break report (the backend cannot derive it).
+            Assert.NotNull(reportedAgeHours);
+            Assert.InRange(reportedAgeHours.Value, 99.9, 100.1);
             // Marker must have been written so the next restart exits cleanly.
             Assert.True(File.Exists(Path.Combine(stateDir, "enrollment-complete.marker")));
             // Session must have been cleared.

src/Agent/AutopilotMonitor.Agent.V2.Core.Tests/Security/SessionIdPersistenceTests.cs

@@ -3,6 +3,7 @@ using System.IO;
 using AutopilotMonitor.Agent.V2.Core.Logging;
 using AutopilotMonitor.Agent.V2.Core.Security;
 using AutopilotMonitor.Agent.V2.Core.Tests.Harness;
+using AutopilotMonitor.Shared.Models;
 using Xunit;
 
 namespace AutopilotMonitor.Agent.V2.Core.Tests.Security
@@ -234,5 +235,78 @@ namespace AutopilotMonitor.Agent.V2.Core.Tests.Security
             Assert.False(File.Exists(Path.Combine(tmp.Path, "session.created")));
             Assert.False(File.Exists(Path.Combine(tmp.Path, "whiteglove.complete")));
         }
+
+        // ============================================================ failed-registration record
+
+        private static RegistrationFailureSummary FailedRun(DateTime at, string outcome = "Failed", bool? linkUpAtEnd = false) =>
+            new RegistrationFailureSummary
+            {
+                FailedRuns = 1,
+                FirstFailedAtUtc = at,
+                LastFailedAtUtc = at,
+                Outcome = outcome,
+                LastError = "No such host is known.",
+                NetworkLinkUpAtStart = false,
+                NetworkLinkUpAtEnd = linkUpAtEnd,
+                AttemptWindowSeconds = 91.3,
+                ConfigFetchOutcome = "UsedDefaults",
+            };
+
+        [Fact]
+        public void RecordRegistrationFailure_counts_runs_keeps_the_first_time_and_describes_the_latest_run()
+        {
+            using var tmp = new TempDirectory();
+            var sut = new SessionIdPersistence(tmp.Path);
+            var first = new DateTime(2026, 8, 20, 9, 0, 0, DateTimeKind.Utc);
+            var second = new DateTime(2026, 9, 12, 7, 30, 0, DateTimeKind.Utc);
+
+            Assert.Null(sut.LoadRegistrationFailure());
+            sut.RecordRegistrationFailure(FailedRun(first));
+            sut.RecordRegistrationFailure(FailedRun(second, outcome: "AuthFailed", linkUpAtEnd: true));
+
+            // A fresh instance reads it back from disk, as the next boot's emergency break does.
+            var record = new SessionIdPersistence(tmp.Path).LoadRegistrationFailure();
+            Assert.NotNull(record);
+            Assert.Equal(2, record!.FailedRuns);
+            Assert.Equal(first, record.FirstFailedAtUtc);
+            Assert.Equal(second, record.LastFailedAtUtc);
+            Assert.Equal("AuthFailed", record.Outcome);
+            Assert.True(record.NetworkLinkUpAtEnd);
+            Assert.Equal(91.3, record.AttemptWindowSeconds);
+            Assert.Equal("UsedDefaults", record.ConfigFetchOutcome);
+        }
+
+        [Fact]
+        public void Registration_failure_record_is_cleared_by_success_delete_and_rotate()
+        {
+            using var tmp = new TempDirectory();
+            var sut = new SessionIdPersistence(tmp.Path);
+            sut.GetOrCreate();
+            var at = new DateTime(2026, 8, 20, 9, 0, 0, DateTimeKind.Utc);
+
+            sut.RecordRegistrationFailure(FailedRun(at));
+            sut.ClearRegistrationFailure();
+            Assert.Null(sut.LoadRegistrationFailure());
+
+            sut.RecordRegistrationFailure(FailedRun(at));
+            sut.Rotate();
+            Assert.Null(sut.LoadRegistrationFailure());
+
+            sut.RecordRegistrationFailure(FailedRun(at));
+            sut.Delete();
+            Assert.Null(sut.LoadRegistrationFailure());
+            Assert.False(File.Exists(Path.Combine(tmp.Path, "registration-failure.json")));
+        }
+
+        [Theory]
+        [InlineData("not json")]
+        [InlineData("{\"FailedRuns\":0}")]
+        public void LoadRegistrationFailure_reports_nothing_for_a_corrupt_or_empty_record(string content)
+        {
+            using var tmp = new TempDirectory();
+            File.WriteAllText(Path.Combine(tmp.Path, "registration-failure.json"), content);
+
+            Assert.Null(new SessionIdPersistence(tmp.Path).LoadRegistrationFailure());
+        }
     }
 }

src/Agent/AutopilotMonitor.Agent.V2.Core.Tests/Security/SessionRegistrationHelperTests.cs

@@ -137,6 +137,86 @@ namespace AutopilotMonitor.Agent.V2.Core.Tests.Security
             Assert.Equal(originalSessionId, result.RotatedFromSessionId);
         }
 
+        private static BackendRefusalException TenantConflict()
+            => new BackendRefusalException("Backend returned 409 Conflict with error code session_owner_mismatch.", 409,
+                AutopilotMonitor.Shared.Constants.AgentErrorCodes.SessionOwnerMismatch);
+
+        [Fact]
+        public async Task Tenant_conflict_409_rotates_once_like_an_owner_mismatch()
+        {
+            using var tmp = new TempDirectory();
+            var logger = NewLogger(tmp);
+            var config = NewConfig();
+            var originalSessionId = config.SessionId;
+            var rotatedTo = Guid.NewGuid().ToString();
+            var seenSessionIds = new System.Collections.Generic.List<string>();
+
+            var apiClient = new FakeApiClient(
+                attempt => attempt == 1
+                    ? throw TenantConflict()
+                    : new RegisterSessionResponse { Success = true, SessionId = rotatedTo, ValidatedBy = ValidatorType.AutopilotV1 },
+                onRegister: r => seenSessionIds.Add(r.SessionId));
+
+            var tracker = new AuthFailureTracker(maxFailures: 5, timeoutMinutes: 0,
+                clock: SystemClock.Instance, logger: logger);
+
+            var result = await SessionRegistrationHelper.RegisterWithRetryAsync(
+                apiClient, config, agentVersion: "0.0.0", logger: logger,
+                authFailureTracker: tracker,
+                backoffDelay: _ => throw new InvalidOperationException("rotation must retry without backoff"),
+                networkLinkWait: _ => Task.CompletedTask,
+                rotateSession: () => { config.SessionId = rotatedTo; return rotatedTo; });
+
+            Assert.Equal(SessionRegistrationOutcome.Succeeded, result.Outcome);
+            Assert.Equal(new[] { originalSessionId, rotatedTo }, seenSessionIds);
+            Assert.Equal(originalSessionId, result.RotatedFromSessionId);
+            Assert.Equal(0, tracker.ConsecutiveFailures);
+        }
+
+        [Fact]
+        public async Task Coded_refusal_after_rotation_fails_without_retry_or_auth_failure()
+        {
+            using var tmp = new TempDirectory();
+            var logger = NewLogger(tmp);
+            var config = NewConfig();
+
+            var apiClient = new FakeApiClient(_ => throw TenantConflict());
+            var tracker = new AuthFailureTracker(maxFailures: 5, timeoutMinutes: 0,
+                clock: SystemClock.Instance, logger: logger);
+
+            var result = await SessionRegistrationHelper.RegisterWithRetryAsync(
+                apiClient, config, agentVersion: "0.0.0", logger: logger,
+                authFailureTracker: tracker,
+                backoffDelay: _ => throw new InvalidOperationException("a coded refusal must not be retried"),
+                networkLinkWait: _ => Task.CompletedTask,
+                rotateSession: () => { config.SessionId = Guid.NewGuid().ToString(); return config.SessionId; });
+
+            Assert.Equal(SessionRegistrationOutcome.Failed, result.Outcome);
+            Assert.Equal(2, apiClient.CallCount);
+            Assert.Equal(0, tracker.ConsecutiveFailures);
+        }
+
+        [Fact]
+        public async Task Rotation_hook_failure_on_a_tenant_conflict_fails_without_auth_failure()
+        {
+            using var tmp = new TempDirectory();
+            var logger = NewLogger(tmp);
+            var config = NewConfig();
+
+            var apiClient = new FakeApiClient(_ => throw TenantConflict());
+            var tracker = new AuthFailureTracker(maxFailures: 5, timeoutMinutes: 0,
+                clock: SystemClock.Instance, logger: logger);
+
+            var result = await SessionRegistrationHelper.RegisterWithRetryAsync(
+                apiClient, config, agentVersion: "0.0.0", logger: logger,
+                authFailureTracker: tracker, networkLinkWait: _ => Task.CompletedTask,
+                rotateSession: () => throw new IOException("disk full"));
+
+            Assert.Equal(SessionRegistrationOutcome.Failed, result.Outcome);
+            Assert.Equal(1, apiClient.CallCount);
+            Assert.Equal(0, tracker.ConsecutiveFailures);
+        }
+
         [Fact]
         public async Task Second_owner_mismatch_after_rotation_is_an_auth_failure()
         {

src/Agent/AutopilotMonitor.Agent.V2.Core.Tests/SignalAdapters/EspAndHelloTrackerAdapterTests.cs

@@ -99,6 +99,52 @@ namespace AutopilotMonitor.Agent.V2.Core.Tests.SignalAdapters
             Assert.Equal(DecisionSignalKind.WhiteGloveShellCoreSuccess, posted.Kind);
         }
 
+        [Theory]
+        [InlineData(true, "true")]
+        [InlineData(false, "false")]
+        public void WhiteGloveEvent_stamps_read_AccountSetup_progress_on_payload_and_evidence(bool progress, string expected)
+        {
+            using var f = new Fixture();
+            using var adapter = new EspAndHelloTrackerAdapter(
+                f.Coordinator, f.Ingress, f.Clock, accountSetupProgressProbe: () => progress);
+
+            adapter.TriggerWhiteGloveFromTest();
+
+            var posted = Assert.Single(f.Ingress.Posted);
+            Assert.Equal(expected, posted.Payload![SignalPayloadKeys.AccountSetupProgress]);
+            Assert.Equal(expected, posted.Evidence.DerivationInputs![SignalPayloadKeys.AccountSetupProgress]);
+        }
+
+        [Fact]
+        public void WhiteGloveEvent_with_unread_AccountSetup_registry_sends_no_progress_key()
+        {
+            // Unknown must keep the sealing classifier's AccountSetup excluder — the payload key
+            // is absent, the evidence records why.
+            using var f = new Fixture();
+            using var adapter = new EspAndHelloTrackerAdapter(
+                f.Coordinator, f.Ingress, f.Clock, accountSetupProgressProbe: () => null);
+
+            adapter.TriggerWhiteGloveFromTest();
+
+            var posted = Assert.Single(f.Ingress.Posted);
+            Assert.True(posted.Payload == null || !posted.Payload.ContainsKey(SignalPayloadKeys.AccountSetupProgress));
+            Assert.Equal("unread", posted.Evidence.DerivationInputs![SignalPayloadKeys.AccountSetupProgress]);
+        }
+
+        [Fact]
+        public void WhiteGloveEvent_default_probe_on_unstarted_coordinator_reports_unread()
+        {
+            // Production probe: the provisioning tracker does not exist before Start (and never on
+            // Device Preparation) — that is "unread", not "untouched".
+            using var f = new Fixture();
+            using var adapter = new EspAndHelloTrackerAdapter(f.Coordinator, f.Ingress, f.Clock);
+
+            adapter.TriggerWhiteGloveFromTest();
+
+            var posted = Assert.Single(f.Ingress.Posted);
+            Assert.True(posted.Payload == null || !posted.Payload.ContainsKey(SignalPayloadKeys.AccountSetupProgress));
+        }
+
         [Fact]
         public void EspFailureEvent_emits_EspTerminalFailure_with_merged_subSource_annotation()
         {

src/Agent/AutopilotMonitor.Agent.V2.Core.Tests/Termination/EnrollmentTerminationHandlerTests.cs

@@ -564,6 +564,37 @@ namespace AutopilotMonitor.Agent.V2.Core.Tests.Termination
             Assert.False(File.Exists(Path.Combine(rig.StateDir, "enrollment-complete.marker")));
         }
 
+        [Fact]
+        public void Handle_whiteglove_part1_writes_marker_before_the_later_exit_steps()
+        {
+            // The technician's reseal can kill the process anywhere after the seal. The marker
+            // is what lets the next boot skip the 48-hour break and resume Part 2, so it must
+            // exist before collectors stop, diagnostics upload, grace and drain run.
+            using var rig = new Rig();
+            rig.State = new DecisionStateBuilder(DecisionState.CreateInitial("S1", "T1")) { Stage = SessionStage.WhiteGloveSealed }.Build();
+            bool? markerPresentAtCollectorStop = null;
+            rig.StopPeripheralCollectorsHook = () => markerPresentAtCollectorStop = rig.SessionPersistence.IsWhiteGloveResume();
+
+            rig.Build().Handle(sender: null!,
+                Args(EnrollmentTerminationReason.DecisionTerminalStage, EnrollmentTerminationOutcome.Succeeded, SessionStage.WhiteGloveSealed));
+
+            Assert.True(markerPresentAtCollectorStop, "whiteglove.complete must be written before the peripheral collectors are stopped.");
+            var emitted = rig.EmittedEventTypes.ToList();
+            Assert.True(emitted.IndexOf("agent_shutting_down") >= 0, "agent_shutting_down must still be emitted.");
+        }
+
+        [Fact]
+        public void Handle_non_whiteglove_termination_never_writes_the_marker()
+        {
+            using var rig = new Rig();
+            rig.State = new DecisionStateBuilder(DecisionState.CreateInitial("S1", "T1")) { Stage = SessionStage.Completed }.Build();
+
+            rig.Build().Handle(sender: null!,
+                Args(EnrollmentTerminationReason.DecisionTerminalStage, EnrollmentTerminationOutcome.Succeeded, SessionStage.Completed));
+
+            Assert.False(rig.SessionPersistence.IsWhiteGloveResume());
+        }
+
         [Fact]
         public void Handle_whiteglove_part1_stops_collectors_before_emitting_part1_complete()
         {

src/Agent/AutopilotMonitor.Agent.V2.Core.Tests/Transport/BackendApiClientAuthClassificationTests.cs

@@ -88,6 +88,35 @@ namespace AutopilotMonitor.Agent.V2.Core.Tests.Transport
             Assert.Null(ex.ErrorCode);
         }
 
+        // ── coded 409: a refusal, never an auth failure ─────────────────────
+
+        [Fact]
+        public async Task Coded_409_is_a_refusal_not_an_auth_failure()
+        {
+            using var client = BuildClient(
+                HttpStatusCode.Conflict,
+                "{\"success\":false,\"message\":\"Session id is already registered to another tenant\",\"errorCode\":\"session_owner_mismatch\"}",
+                "application/json");
+
+            var ex = await Assert.ThrowsAsync<BackendRefusalException>(
+                () => client.RegisterSessionAsync(new AutopilotMonitor.Shared.Models.SessionRegistration()));
+
+            Assert.Equal(409, ex.StatusCode);
+            Assert.Equal("session_owner_mismatch", ex.ErrorCode);
+        }
+
+        [Fact]
+        public async Task Uncoded_409_stays_a_plain_http_error()
+        {
+            using var client = BuildClient(
+                HttpStatusCode.Conflict,
+                "{\"success\":false,\"message\":\"A diagnostics package with this name already exists\"}",
+                "application/json");
+
+            await Assert.ThrowsAsync<HttpRequestException>(
+                () => client.GetDiagnosticsUploadUrlAsync("tenant-1", "session-1", "package.zip"));
+        }
+
         [Theory]
         [InlineData(null, null)]
         [InlineData("", null)]

src/Agent/AutopilotMonitor.Agent.V2.Core/Configuration/AgentConfiguration.cs

@@ -157,12 +157,13 @@ namespace AutopilotMonitor.Agent.V2.Core.Configuration
         public double ReplaySpeedFactor { get; set; } = 50;
 
         /// <summary>
-        /// Wait time in seconds after ESP exit before marking Hello as skipped.
-        /// When ESP exits, we wait this duration for Hello wizard to start (event 62404).
-        /// Default: 30 seconds (reasonable for systems under load).
-        /// If Hello wizard starts within this window, we continue waiting for Hello completion (300/301).
+        /// Tenant Hello wait in seconds after the ESP exit, merged from RemoteConfig. Stamped on
+        /// <c>EnrollmentFactsObserved</c> for the decision engine's hello_safety window; the Hello
+        /// tracker gets the same remote value through the collector configuration. Both resolve it
+        /// through <see cref="HelloWaitTimeout.EffectiveSeconds"/> (at least the built-in 300 s,
+        /// at most 3600 s). Default: 300 seconds.
         /// </summary>
-        public int HelloWaitTimeoutSeconds { get; set; } = 30;
+        public int HelloWaitTimeoutSeconds { get; set; } = HelloWaitTimeout.DefaultSeconds;
 
         /// <summary>
         /// Maximum consecutive authentication failures (401/403) before the agent shuts down.

src/Agent/AutopilotMonitor.Agent.V2.Core/Monitoring/Enrollment/SystemSignals/EspAndHelloTracker.cs

@@ -216,6 +216,13 @@ namespace AutopilotMonitor.Agent.V2.Core.Monitoring.Enrollment.SystemSignals
         /// </summary>
         public bool HasAccountSetupProgress => _provisioningTracker?.HasAccountSetupProgress ?? false;
 
+        /// <summary>
+        /// <see cref="ProvisioningStatusTracker.AccountSetupProgressIfRead"/>; <c>null</c> also when
+        /// the provisioning tracker is not running (Device Preparation, before Start). Stamped on
+        /// the WhiteGloveShellCoreSuccess signal for the sealing classifier.
+        /// </summary>
+        public bool? AccountSetupProgressIfRead => _provisioningTracker?.AccountSetupProgressIfRead;
+
         /// <summary>
         /// True once WhiteGlove start has been detected (EventID 509 or persisted from prior run).
         /// Forwarded from ModernDeploymentTracker.
@@ -234,7 +241,7 @@ namespace AutopilotMonitor.Agent.V2.Core.Monitoring.Enrollment.SystemSignals
             string tenantId,
             InformationalEventPost post,
             AgentLogger logger,
-            int helloWaitTimeoutSeconds = 30,
+            int helloWaitTimeoutSeconds = HelloWaitTimeout.DefaultSeconds,
             bool modernDeploymentWatcherEnabled = true,
             int modernDeploymentLogLevelMax = 3,
             bool modernDeploymentBackfillEnabled = true,

src/Agent/AutopilotMonitor.Agent.V2.Core/Monitoring/Enrollment/SystemSignals/HelloTracker.cs

@@ -47,7 +47,6 @@ namespace AutopilotMonitor.Agent.V2.Core.Monitoring.Enrollment.SystemSignals
 
         internal const string HResult_UserSkippedHello = "0x801C044F";
 
-        internal const int HelloCompletionTimeoutSeconds = 300;
         internal const int BackfillLookbackMinutes = 5;
 
         private static readonly HashSet<int> TrackedUdrEventIds = new HashSet<int>
@@ -81,6 +80,12 @@ namespace AutopilotMonitor.Agent.V2.Core.Monitoring.Enrollment.SystemSignals
         private readonly InformationalEventPost _post;
         private readonly int _helloWaitTimeoutSeconds;
 
+        // How long a started Hello wizard may take to reach a terminal event (300/301/362): the
+        // tenant's Hello wait resolved through HelloWaitTimeout (300..3600 s). The engine's
+        // hello_safety window spends the same budget from the ESP exit, so this timer never cuts
+        // a user off inside the wizard before the tenant's budget is used up.
+        private readonly int _helloCompletionTimeoutSeconds;
+
         private EventLogWatcher _udrWatcher;
         private EventLogWatcher _helloForBusinessWatcher;
         private System.Threading.Timer _policyCheckTimer;
@@ -121,13 +126,14 @@ namespace AutopilotMonitor.Agent.V2.Core.Monitoring.Enrollment.SystemSignals
             string tenantId,
             InformationalEventPost post,
             AgentLogger logger,
-            int helloWaitTimeoutSeconds = 30)
+            int helloWaitTimeoutSeconds = HelloWaitTimeout.DefaultSeconds)
         {
             _sessionId = sessionId ?? throw new ArgumentNullException(nameof(sessionId));
             _tenantId = tenantId ?? throw new ArgumentNullException(nameof(tenantId));
             _post = post ?? throw new ArgumentNullException(nameof(post));
             _logger = logger ?? throw new ArgumentNullException(nameof(logger));
             _helloWaitTimeoutSeconds = helloWaitTimeoutSeconds;
+            _helloCompletionTimeoutSeconds = HelloWaitTimeout.EffectiveSeconds(helloWaitTimeoutSeconds);
         }
 
         // =====================================================================
@@ -1094,7 +1100,7 @@ namespace AutopilotMonitor.Agent.V2.Core.Monitoring.Enrollment.SystemSignals
                 if (policyKnownEnabled)
                 {
                     _logger.Info($"Hello wait timeout ({_helloWaitTimeoutSeconds}s) expired but Hello policy is enabled — " +
-                                 $"wizard not yet visible, starting long completion timer ({HelloCompletionTimeoutSeconds}s)");
+                                 $"wizard not yet visible, starting long completion timer ({_helloCompletionTimeoutSeconds}s)");
 
                     _post.Emit(new EnrollmentEvent
                     {
@@ -1105,7 +1111,7 @@ namespace AutopilotMonitor.Agent.V2.Core.Monitoring.Enrollment.SystemSignals
                         Source = "EspAndHelloTracker",
                         Phase = EnrollmentPhase.Unknown,
                         Message = $"Hello wizard did not start within {_helloWaitTimeoutSeconds}s after ESP exit — " +
-                                  $"Hello policy is enabled, waiting up to {HelloCompletionTimeoutSeconds}s for wizard",
+                                  $"Hello policy is enabled, waiting up to {_helloCompletionTimeoutSeconds}s for wizard",
                         Data = new Dictionary<string, object>
                         {
                             { "timeoutSeconds", _helloWaitTimeoutSeconds },
@@ -1205,11 +1211,11 @@ namespace AutopilotMonitor.Agent.V2.Core.Monitoring.Enrollment.SystemSignals
             if (_isHelloCompleted) return;
             if (_helloCompletionTimer != null) return;
 
-            _logger.Info($"Starting Hello completion timer ({HelloCompletionTimeoutSeconds}s) - waiting for terminal Hello event (300/301/362)");
+            _logger.Info($"Starting Hello completion timer ({_helloCompletionTimeoutSeconds}s) - waiting for terminal Hello event (300/301/362)");
             _helloCompletionTimer = new System.Threading.Timer(
                 OnHelloCompletionTimeout,
                 null,
-                TimeSpan.FromSeconds(HelloCompletionTimeoutSeconds),
+                TimeSpan.FromSeconds(_helloCompletionTimeoutSeconds),
                 TimeSpan.FromMilliseconds(-1));
         }
 
@@ -1230,7 +1236,7 @@ namespace AutopilotMonitor.Agent.V2.Core.Monitoring.Enrollment.SystemSignals
                     return;
                 }
 
-                _logger.Warning($"Hello completion timeout ({HelloCompletionTimeoutSeconds}s) expired after wizard start without terminal event");
+                _logger.Warning($"Hello completion timeout ({_helloCompletionTimeoutSeconds}s) expired after wizard start without terminal event");
                 _isHelloCompleted = true;
                 HelloOutcome = _helloWizardStarted ? "timeout" : "wizard_not_started";
                 StopHelloCompletionTimerLocked();
@@ -1243,10 +1249,10 @@ namespace AutopilotMonitor.Agent.V2.Core.Monitoring.Enrollment.SystemSignals
                     Severity = EventSeverity.Warning,
                     Source = "EspAndHelloTracker",
                     Phase = EnrollmentPhase.Unknown,
-                    Message = $"Hello wizard started but no terminal event (300/301/362) arrived within {HelloCompletionTimeoutSeconds}s",
+                    Message = $"Hello wizard started but no terminal event (300/301/362) arrived within {_helloCompletionTimeoutSeconds}s",
                     Data = new Dictionary<string, object>
                     {
-                        { "timeoutSeconds", HelloCompletionTimeoutSeconds },
+                        { "timeoutSeconds", _helloCompletionTimeoutSeconds },
                         { "helloWizardStarted", _helloWizardStarted }
                     },
                     ImmediateUpload = true
@@ -1304,6 +1310,7 @@ namespace AutopilotMonitor.Agent.V2.Core.Monitoring.Enrollment.SystemSignals
         internal void TriggerWaitTimeoutForTest() => OnHelloWaitTimeout(null);
         internal void TriggerCompletionTimeoutForTest() => OnHelloCompletionTimeout(null);
 
+        internal int CompletionTimeoutSecondsForTest => _helloCompletionTimeoutSeconds;
         internal bool IsWaitTimerActiveForTest { get { lock (_stateLock) { return _helloWaitTimer != null; } } }
         internal bool IsCompletionTimerActiveForTest { get { lock (_stateLock) { return _helloCompletionTimer != null; } } }
         internal bool IsHelloWizardStartedForTest { get { lock (_stateLock) { return _helloWizardStarted; } } }

src/Agent/AutopilotMonitor.Agent.V2.Core/Monitoring/Enrollment/SystemSignals/ProvisioningStatusTracker.cs

@@ -254,16 +254,27 @@ namespace AutopilotMonitor.Agent.V2.Core.Monitoring.Enrollment.SystemSignals
         /// host to tell the real user-ESP page exit from the Device→Account handoff exit; not a
         /// completion input.
         /// </summary>
-        public bool HasAccountSetupProgress
+        public bool HasAccountSetupProgress => AccountSetupProgressIfRead == true;
+
+        /// <summary>
+        /// Tri-state form of <see cref="HasAccountSetupProgress"/>: <c>null</c> until the tracker
+        /// has actually read AccountSetup subcategories (not started yet, category not written,
+        /// empty), otherwise whether one has left <c>notStarted</c>. The WhiteGlove sealing
+        /// classifier treats only a read <c>false</c> as "user ESP untouched" — an unread
+        /// registry must never pass for an untouched one.
+        /// </summary>
+        public bool? AccountSetupProgressIfRead
         {
             get
             {
                 lock (_stateLock)
                 {
                     if (_lastSubcategoryStates == null)
-                        return false;
-                    if (!_lastSubcategoryStates.TryGetValue("AccountSetupCategory.Status", out var subs) || subs == null)
-                        return false;
+                        return null;
+                    if (!_lastSubcategoryStates.TryGetValue("AccountSetupCategory.Status", out var subs)
+                        || subs == null
+                        || subs.Count == 0)
+                        return null;
                     foreach (var state in subs.Values)
                     {
                         if (!string.Equals(state, "notStarted", StringComparison.OrdinalIgnoreCase))
@@ -279,7 +290,7 @@ namespace AutopilotMonitor.Agent.V2.Core.Monitoring.Enrollment.SystemSignals
         /// with subcategoryState=succeeded. NOT a WhiteGlove indicator: on Win11 25H2 (build
         /// 26200+) Windows writes this property as a trivially-succeeding step in EVERY
         /// enrollment (validated 2026-08-18 — user-driven sessions f475e697/c601a24b/4219203f
-        /// all carry it). Kept for observability only (whiteglove_signal_* raw dumps); must not
+        /// all carry it). Kept for observability only (save_result_step_* raw dumps); must not
         /// be used for classification. Reliable WhiteGlove anchors are ModernDeployment
         /// EventID 509 and the Shell-Core WhiteGlove-success events.
         /// </summary>
@@ -966,8 +977,10 @@ namespace AutopilotMonitor.Agent.V2.Core.Monitoring.Enrollment.SystemSignals
                                     _logger.Info($"ProvisioningStatusTracker: SaveWhiteGloveSuccessResult state " +
                                                  $"transition: {previousState} -> {state}");
 
+                                    // Neutral trigger name: the step exists in every 25H2 enrollment,
+                                    // and "whiteglove_*" was read as a pre-provisioning signal.
                                     EmitRawRegistryDump(categoryLabel, jsonValue,
-                                        $"whiteglove_signal_{state}");
+                                        $"save_result_step_{state}");
 
                                     if (string.Equals(state, "succeeded", StringComparison.OrdinalIgnoreCase))
                                     {

src/Agent/AutopilotMonitor.Agent.V2.Core/Monitoring/Enrollment/SystemSignals/StallProbeCollector.cs

@@ -39,7 +39,8 @@ namespace AutopilotMonitor.Agent.V2.Core.Monitoring.Enrollment.SystemSignals
     ///   - Trace-indexed probes: emit a <c>stall_probe_check</c> Trace event on quiet/activeInstalls.
     ///   - Non-trace probes: emit nothing when no anomaly is found (agent log only).
     ///   - Anomaly found: always emit <c>stall_probe_result</c> Warning (plus EspFailureDetected if terminal).
-    ///   - Probe index configured as "session-stalled": emit fire-once <c>session_stalled</c> Warning.
+    ///   - Probe index configured as "session-stalled": emit fire-once <c>session_stalled</c> Warning
+    ///     (held back while a tenant-extended Hello wait is pending, emitted once it is over).
     /// </summary>
     public class StallProbeCollector
     {
@@ -54,10 +55,20 @@ namespace AutopilotMonitor.Agent.V2.Core.Monitoring.Enrollment.SystemSignals
         private readonly HashSet<int> _harmlessModernDeploymentEventIds;
         private readonly bool _isDevicePreparation;
 
+        // True while a tenant-extended Hello wait (up to 60 min) is pending in the decision
+        // engine. It is a bounded wait that resolves the session either way — not a stall — so
+        // session_stalled is held back until it is over.
+        private readonly Func<bool> _helloWaitPending;
+
         private readonly object _stateLock = new object();
         private readonly HashSet<int> _firedProbeIndices = new HashSet<int>();
         private bool _sessionStalledFired;
 
+        // The session_stalled report held back while the Hello wait was pending (the probe result
+        // of that idle window). Emitted on a later tick once the wait is over and the session is
+        // still idle; dropped by real activity (ResetProbes).
+        private ProbeResult _deferredSessionStalled;
+
         // P2 — monotonic generation, bumped by ResetProbes. RunProbe runs OUTSIDE _stateLock; if a
         // real-activity reset lands while a probe is in flight, the post-probe "mark fired" must be
         // dropped (its generation is stale) so the next idle period can fire that probe again.
@@ -153,9 +164,11 @@ namespace AutopilotMonitor.Agent.V2.Core.Monitoring.Enrollment.SystemSignals
             string[] sources,
             int sessionStalledAfterProbeIndex,
             int[] harmlessModernDeploymentEventIds = null,
-            bool isDevicePreparation = false)
+            bool isDevicePreparation = false,
+            Func<bool> helloWaitPending = null)
         {
             _sessionId = sessionId ?? throw new ArgumentNullException(nameof(sessionId));
+            _helloWaitPending = helloWaitPending;
             _tenantId = tenantId ?? throw new ArgumentNullException(nameof(tenantId));
             _post = post ?? throw new ArgumentNullException(nameof(post));
             _logger = logger ?? throw new ArgumentNullException(nameof(logger));
@@ -192,6 +205,7 @@ namespace AutopilotMonitor.Agent.V2.Core.Monitoring.Enrollment.SystemSignals
                 // Always bump the generation, even when no probes are currently marked fired — a
                 // probe may be running right now (outside the lock) and must not re-mark itself.
                 _probeGeneration++;
+                _deferredSessionStalled = null;
                 if (_firedProbeIndices.Count > 0)
                 {
                     _logger.Trace($"StallProbeCollector: resetting {_firedProbeIndices.Count} fired probes");
@@ -206,6 +220,8 @@ namespace AutopilotMonitor.Agent.V2.Core.Monitoring.Enrollment.SystemSignals
         /// </summary>
         public void CheckAndRunProbes(double idleMinutes)
         {
+            EmitDeferredSessionStalledIfDue(idleMinutes);
+
             // Iterate thresholds from lowest to highest so probes fire in order.
             for (int i = 0; i < _thresholdsMinutes.Length; i++)
             {
@@ -225,7 +241,7 @@ namespace AutopilotMonitor.Agent.V2.Core.Monitoring.Enrollment.SystemSignals
                 if (alreadyFired)
                     continue;
 
-                RunProbe(probeIndex, thresholdMinutes, idleMinutes);
+                RunProbe(probeIndex, thresholdMinutes, idleMinutes, genAtStart);
 
                 lock (_stateLock)
                 {
@@ -237,7 +253,7 @@ namespace AutopilotMonitor.Agent.V2.Core.Monitoring.Enrollment.SystemSignals
             }
         }
 
-        private void RunProbe(int probeIndex, int thresholdMinutes, double idleMinutes)
+        private void RunProbe(int probeIndex, int thresholdMinutes, double idleMinutes, int generation)
         {
             var started = DateTime.UtcNow;
             var result = new ProbeResult { ProbeIndex = probeIndex, IdleMinutes = Math.Round(idleMinutes, 1) };
@@ -293,8 +309,21 @@ namespace AutopilotMonitor.Agent.V2.Core.Monitoring.Enrollment.SystemSignals
 
             // session_stalled is fire-once across the whole session. It fires at the configured
             // probe index UNLESS activeInstalls is observed (activeInstalls counts as progress).
+            // While a tenant Hello wait is pending it is held back, not dropped.
             if (probeIndex == _sessionStalledAfterProbeIndex && !hasActiveInstalls)
             {
+                if (IsHelloWaitPending())
+                {
+                    lock (_stateLock)
+                    {
+                        // P2: a reset that landed while this probe ran ended the idle window.
+                        if (!_sessionStalledFired && _probeGeneration == generation)
+                            _deferredSessionStalled = result;
+                    }
+                    _logger.Info($"StallProbeCollector: session_stalled held back at probe {probeIndex} — a tenant Hello wait is still pending");
+                    return;
+                }
+
                 lock (_stateLock)
                 {
                     if (_sessionStalledFired)
@@ -305,6 +334,52 @@ namespace AutopilotMonitor.Agent.V2.Core.Monitoring.Enrollment.SystemSignals
             }
         }
 
+        /// <summary>
+        /// Emits the held-back session_stalled report once the Hello wait is over and the session
+        /// is still idle (no ResetProbes since), with the idle time of this tick.
+        /// </summary>
+        private void EmitDeferredSessionStalledIfDue(double idleMinutes)
+        {
+            ProbeResult deferred;
+            lock (_stateLock)
+            {
+                deferred = _deferredSessionStalled;
+                if (deferred == null || _sessionStalledFired)
+                    return;
+            }
+
+            if (IsHelloWaitPending())
+                return;
+
+            lock (_stateLock)
+            {
+                if (_sessionStalledFired || !ReferenceEquals(_deferredSessionStalled, deferred))
+                    return;
+                _sessionStalledFired = true;
+                _deferredSessionStalled = null;
+            }
+
+            deferred.IdleMinutes = Math.Round(idleMinutes, 1);
+            _logger.Info($"StallProbeCollector: Hello wait over, still idle at {deferred.IdleMinutes}min — emitting the held-back session_stalled");
+            EmitSessionStalledEvent(deferred);
+        }
+
+        private bool IsHelloWaitPending()
+        {
+            if (_helloWaitPending == null)
+                return false;
+            try
+            {
+                return _helloWaitPending();
+            }
+            catch (Exception ex)
+            {
+                // A failing probe must never cost the stall report — treat as not pending.
+                _logger.Warning($"StallProbeCollector: Hello-wait probe threw, treating as not pending: {ex.Message}");
+                return false;
+            }
+        }
+
         private void RunWithTimeout(Action<ProbeResult> scan, int timeoutMs, string sourceName, ProbeResult result)
         {
             // Enforce a real hard timeout: run the source on a dedicated background thread and

src/Agent/AutopilotMonitor.Agent.V2.Core/Monitoring/Transport/BackendApiClient.cs

@@ -237,6 +237,7 @@ namespace AutopilotMonitor.Agent.V2.Core.Monitoring.Transport
                 using (response)
                 {
                     await ThrowOnAuthFailureAsync(response).ConfigureAwait(false);
+                    await ThrowOnCodedConflictAsync(response).ConfigureAwait(false);
                     response.EnsureSuccessStatusCode();
 
                     var responseJson = await response.Content.ReadAsStringAsync().ConfigureAwait(false);
@@ -338,6 +339,38 @@ namespace AutopilotMonitor.Agent.V2.Core.Monitoring.Transport
                 errorCode: errorCode);
         }
 
+        /// <summary>
+        /// Throws <see cref="BackendRefusalException"/> for a 409 whose JSON body carries a
+        /// machine-readable <c>errorCode</c> — register-session answers
+        /// <c>session_owner_mismatch</c> when another tenant claimed the session id, which no retry
+        /// can clear but a session rotation can. A 409 without a code falls through to
+        /// <c>EnsureSuccessStatusCode</c> as before.
+        /// </summary>
+        private static async Task ThrowOnCodedConflictAsync(HttpResponseMessage response)
+        {
+            if (response.StatusCode != System.Net.HttpStatusCode.Conflict)
+                return;
+
+            string body = null;
+            try
+            {
+                body = await response.Content.ReadAsStringAsync().ConfigureAwait(false);
+            }
+            catch
+            {
+                // No body, no code — the generic 409 handling applies.
+            }
+
+            var errorCode = TryExtractErrorCode(body);
+            if (errorCode == null)
+                return;
+
+            throw new BackendRefusalException(
+                $"Backend returned 409 Conflict with error code {errorCode}.",
+                (int)response.StatusCode,
+                errorCode);
+        }
+
         /// <summary>
         /// Pulls <c>errorCode</c> out of a backend JSON error body. Tolerant: any parse failure
         /// or a non-string value yields null, never an exception.
@@ -471,4 +504,24 @@ namespace AutopilotMonitor.Agent.V2.Core.Monitoring.Transport
             ErrorCode = errorCode;
         }
     }
+
+    /// <summary>
+    /// Thrown when the backend refuses a request on a non-auth status (409) with a machine-readable
+    /// code from Shared <c>Constants.AgentErrorCodes</c>. Unlike <see cref="BackendAuthException"/>
+    /// it says nothing about the device's authorization and must never feed the auth-failure tracker.
+    /// </summary>
+    public class BackendRefusalException : Exception
+    {
+        /// <summary>HTTP status code returned by the backend (409).</summary>
+        public int StatusCode { get; }
+
+        /// <summary>Machine-readable code from the backend's JSON body (never null).</summary>
+        public string ErrorCode { get; }
+
+        public BackendRefusalException(string message, int statusCode, string errorCode) : base(message)
+        {
+            StatusCode = statusCode;
+            ErrorCode = errorCode;
+        }
+    }
 }

src/Agent/AutopilotMonitor.Agent.V2.Core/Monitoring/Transport/EmergencyReporter.cs

@@ -79,6 +79,11 @@ namespace AutopilotMonitor.Agent.V2.Core.Monitoring.Transport
         /// exist for callers whose process is about to exit and cannot come back later
         /// (emergency break): retries stay INSIDE the one reservation, so they never
         /// multiply against the per-session budget.
+        ///
+        /// <paramref name="sessionAgeHours"/> is the emergency break's measured session age
+        /// (<see cref="AgentErrorReport.SessionAgeHours"/>) and <paramref name="priorRegistrationFailure"/>
+        /// its record of earlier starts that never registered
+        /// (<see cref="AgentErrorReport.PriorRegistrationFailure"/>); null for every other error type.
         /// </summary>
         public virtual async Task TrySendAsync(
             AgentErrorType errorType,
@@ -87,7 +92,9 @@ namespace AutopilotMonitor.Agent.V2.Core.Monitoring.Transport
             long? sequenceNumber = null,
             int attempts = 1,
             TimeSpan? perAttemptTimeout = null,
-            TimeSpan? retryDelay = null)
+            TimeSpan? retryDelay = null,
+            double? sessionAgeHours = null,
+            RegistrationFailureSummary priorRegistrationFailure = null)
         {
             // Deduplicate by error type + status code: same failure category sent only once per session.
             // All anti-flood checks are protected by a lock to prevent race conditions when
@@ -135,6 +142,8 @@ namespace AutopilotMonitor.Agent.V2.Core.Monitoring.Transport
                 Message = message,
                 HttpStatusCode = httpStatusCode,
                 SequenceNumber = sequenceNumber,
+                SessionAgeHours = sessionAgeHours,
+                PriorRegistrationFailure = priorRegistrationFailure,
                 AgentVersion = _agentVersion,
                 Timestamp = DateTime.UtcNow
             };

src/Agent/AutopilotMonitor.Agent.V2.Core/Orchestration/DefaultComponentFactory.cs

@@ -479,6 +479,12 @@ namespace AutopilotMonitor.Agent.V2.Core.Orchestration
 
             if (collectors.StallProbeEnabled)
             {
+                // A pending tenant Hello wait (up to 60 min) is a bounded wait that resolves the
+                // session either way — the probe holds session_stalled back until it is over.
+                Func<bool>? helloWaitPending = null;
+                if (decisionStateProbe != null)
+                    helloWaitPending = () => DecisionEngine.IsTenantHelloWaitPending(decisionStateProbe(), clock.UtcNow);
+
                 hosts.Add(new StallProbeHost(
                     sessionId: sessionId,
                     tenantId: tenantId,
@@ -490,7 +496,8 @@ namespace AutopilotMonitor.Agent.V2.Core.Orchestration
                     sources: collectors.StallProbeSources,
                     sessionStalledAfterProbeIndex: collectors.SessionStalledAfterProbeIndex,
                     harmlessModernDeploymentEventIds: collectors.ModernDeploymentHarmlessEventIds,
-                    isDevicePreparation: isDevicePreparation));
+                    isDevicePreparation: isDevicePreparation,
+                    helloWaitPending: helloWaitPending));
             }
 
             // ESP policy-provider stall tripwire — always-on kernel host (no config gate, like

src/Agent/AutopilotMonitor.Agent.V2.Core/Orchestration/Hosts/StallProbeHost.cs

@@ -58,7 +58,8 @@ namespace AutopilotMonitor.Agent.V2.Core.Orchestration
             string[]? sources,
             int sessionStalledAfterProbeIndex,
             int[]? harmlessModernDeploymentEventIds,
-            bool isDevicePreparation = false)
+            bool isDevicePreparation = false,
+            Func<bool>? helloWaitPending = null)
         {
             if (ingress == null) throw new ArgumentNullException(nameof(ingress));
             if (clock == null) throw new ArgumentNullException(nameof(clock));
@@ -74,7 +75,8 @@ namespace AutopilotMonitor.Agent.V2.Core.Orchestration
                 sources: sources ?? new[] { "provisioning_registry", "diagnostics_registry", "eventlog", "appworkload_log" },
                 sessionStalledAfterProbeIndex: sessionStalledAfterProbeIndex,
                 harmlessModernDeploymentEventIds: harmlessModernDeploymentEventIds,
-                isDevicePreparation: isDevicePreparation);
+                isDevicePreparation: isDevicePreparation,
+                helloWaitPending: helloWaitPending);
 
             _adapter = new StallProbeCollectorAdapter(_collector, ingress, clock);
             _observableIngress = ingress as SignalIngress;

src/Agent/AutopilotMonitor.Agent.V2.Core/Orchestration/RecoveryCoordinator.cs

@@ -420,7 +420,12 @@ namespace AutopilotMonitor.Agent.V2.Core.Orchestration
                 // Re-stamp AgentBootUtc so the current run's deadlines floor at "now" — the
                 // persisted boot anchor is from the prior run and using it would let replayed
                 // tail signals arm deadlines that are already past-due (premature fire).
-                seed = loadedState.ToBuilder().WithAgentBootUtc(agentBootUtc).Build();
+                // Re-stamp SessionId too: after a session rotation at registration
+                // (SESSION-OWNER-BINDING) the snapshot still names the refused session, and the
+                // timeline emitter takes each event's SessionId from this state.
+                var seedBuilder = loadedState.ToBuilder().WithAgentBootUtc(agentBootUtc);
+                seedBuilder.SessionId = sessionId;
+                seed = seedBuilder.Build();
                 signalsToReplay = CollectSignalLogTailAfter(signalLog, loadedState.LastAppliedSignalOrdinal);
                 branchTag = signalsToReplay.Count > 0 ? "b-tail-replay" : "b-snapshot-current";
             }

src/Agent/AutopilotMonitor.Agent.V2.Core/Security/SessionIdPersistence.cs

@@ -1,6 +1,8 @@
 using System;
 using System.IO;
 using AutopilotMonitor.Agent.V2.Core.Logging;
+using AutopilotMonitor.Shared.Models;
+using Newtonsoft.Json;
 
 namespace AutopilotMonitor.Agent.V2.Core.Security
 {
@@ -22,6 +24,7 @@ namespace AutopilotMonitor.Agent.V2.Core.Security
         private readonly string _sessionFilePath;
         private readonly string _sessionCreatedFilePath;
         private readonly string _whiteGloveMarkerPath;
+        private readonly string _registrationFailurePath;
         private readonly object _lockObject = new object();
 
         public SessionIdPersistence(string dataDirectory)
@@ -35,6 +38,7 @@ namespace AutopilotMonitor.Agent.V2.Core.Security
             _sessionFilePath = Path.Combine(dataDirectory, "session.id");
             _sessionCreatedFilePath = Path.Combine(dataDirectory, "session.created");
             _whiteGloveMarkerPath = Path.Combine(dataDirectory, "whiteglove.complete");
+            _registrationFailurePath = Path.Combine(dataDirectory, "registration-failure.json");
         }
 
         /// <summary>
@@ -164,6 +168,55 @@ namespace AutopilotMonitor.Agent.V2.Core.Security
             lock (_lockObject) TryDelete(_whiteGloveMarkerPath, logger);
         }
 
+        /// <summary>
+        /// Folds one failed registration run into <c>registration-failure.json</c>: the run
+        /// counter grows, the first failure time is kept, everything else describes the latest
+        /// run. A run that never registers exits and leaves nothing else behind — the emergency
+        /// break reads this file so the backend learns why the session never appeared.
+        /// Best-effort; never throws.
+        /// </summary>
+        public void RecordRegistrationFailure(RegistrationFailureSummary latestRun, AgentLogger logger = null)
+        {
+            if (latestRun == null) return;
+            lock (_lockObject)
+            {
+                try
+                {
+                    var previous = LoadRegistrationFailureInternal();
+                    var merged = new RegistrationFailureSummary
+                    {
+                        FailedRuns = (previous?.FailedRuns ?? 0) + 1,
+                        FirstFailedAtUtc = previous?.FirstFailedAtUtc ?? latestRun.LastFailedAtUtc,
+                        LastFailedAtUtc = latestRun.LastFailedAtUtc,
+                        Outcome = latestRun.Outcome,
+                        LastError = latestRun.LastError,
+                        NetworkLinkUpAtStart = latestRun.NetworkLinkUpAtStart,
+                        NetworkLinkUpAtEnd = latestRun.NetworkLinkUpAtEnd,
+                        AttemptWindowSeconds = latestRun.AttemptWindowSeconds,
+                        ConfigFetchOutcome = latestRun.ConfigFetchOutcome,
+                    };
+                    File.WriteAllText(_registrationFailurePath, JsonConvert.SerializeObject(merged));
+                    logger?.Info($"SessionIdPersistence: recorded failed registration run #{merged.FailedRuns}.");
+                }
+                catch (Exception ex)
+                {
+                    logger?.Warning($"SessionIdPersistence: RecordRegistrationFailure failed: {ex.Message}");
+                }
+            }
+        }
+
+        /// <summary>The failed-registration record since the last success, or <c>null</c> (none, unreadable).</summary>
+        public RegistrationFailureSummary LoadRegistrationFailure()
+        {
+            lock (_lockObject) return LoadRegistrationFailureInternal();
+        }
+
+        /// <summary>Drops the failed-registration record — called once registration succeeds.</summary>
+        public void ClearRegistrationFailure(AgentLogger logger = null)
+        {
+            lock (_lockObject) TryDelete(_registrationFailurePath, logger);
+        }
+
         /// <summary>Deletes the persisted SessionId. Next <see cref="GetOrCreate"/> starts a fresh session.</summary>
         public void Delete(AgentLogger logger = null)
         {
@@ -172,6 +225,7 @@ namespace AutopilotMonitor.Agent.V2.Core.Security
                 TryDelete(_sessionFilePath, logger);
                 TryDelete(_sessionCreatedFilePath, logger);
                 TryDelete(_whiteGloveMarkerPath, logger);
+                TryDelete(_registrationFailurePath, logger);
             }
         }
 
@@ -191,11 +245,28 @@ namespace AutopilotMonitor.Agent.V2.Core.Security
                 var fresh = Guid.NewGuid().ToString();
                 WriteAtomic(fresh, logger);
                 SaveSessionCreatedAtInternal(DateTime.UtcNow);
+                // Earlier failed runs belonged to the refused session id.
+                TryDelete(_registrationFailurePath, logger);
                 logger?.Info($"SessionIdPersistence: rotated SessionId to {fresh}.");
                 return fresh;
             }
         }
 
+        private RegistrationFailureSummary LoadRegistrationFailureInternal()
+        {
+            try
+            {
+                if (!File.Exists(_registrationFailurePath)) return null;
+                var record = JsonConvert.DeserializeObject<RegistrationFailureSummary>(File.ReadAllText(_registrationFailurePath));
+                return record != null && record.FailedRuns > 0 ? record : null;
+            }
+            catch
+            {
+                // Corrupt record: report nothing rather than a half-read account.
+                return null;
+            }
+        }
+
         private static void TryDelete(string path, AgentLogger logger)
         {
             try { if (File.Exists(path)) File.Delete(path); }

src/Agent/AutopilotMonitor.Agent.V2.Core/Security/SessionRegistrationHelper.cs

@@ -34,13 +34,16 @@ namespace AutopilotMonitor.Agent.V2.Core.Security
     ///     <see cref="AuthFailureTracker"/> (which fires the first-failure distress + may trip
     ///     the shutdown threshold) and return <see cref="SessionRegistrationOutcome.AuthFailed"/>
     ///     without retrying — the backend has definitively rejected the device cert.</item>
-    ///   <item>Exception to the above — a 403 carrying error code
+    ///   <item>Exception to the above — a 403 or 409 carrying error code
     ///     <c>session_owner_mismatch</c> (SESSION-OWNER-BINDING): the persisted SessionId names a
-    ///     session bound to another device identity (typically an Intune re-enrollment without a
-    ///     wipe, where <c>session.id</c> survived but the certificate identity changed). That is
-    ///     not an auth failure: the session is rotated through <c>rotateSession</c> and
-    ///     registration retried exactly once. The tracker is NOT fed — five of these would
-    ///     otherwise soft-shutdown a perfectly authorized agent.</item>
+    ///     session that is not this device's — bound to another device identity (403, typically an
+    ///     Intune re-enrollment without a wipe, where <c>session.id</c> survived but the certificate
+    ///     identity changed) or claimed by another tenant (409). That is not an auth failure: the
+    ///     session is rotated through <c>rotateSession</c> and registration retried exactly once.
+    ///     The tracker is NOT fed — five of these would otherwise soft-shutdown a perfectly
+    ///     authorized agent.</item>
+    ///   <item>Any other coded 409 (<see cref="BackendRefusalException"/>) → <see cref="SessionRegistrationOutcome.Failed"/>
+    ///     without retrying — a refusal with a code does not clear by waiting.</item>
     ///   <item>Any other exception on the last attempt → <see cref="EmergencyReporter.TrySendAsync"/>
     ///     with <c>AgentErrorType.RegisterSessionFailed</c> so operators see the final cause.</item>
     /// </list>
@@ -113,14 +116,14 @@ namespace AutopilotMonitor.Agent.V2.Core.Security
                     lastError = response?.Message ?? "(null response)";
                     logger.Warning($"Session registration failed: {lastError}");
                 }
-                catch (BackendAuthException ex) when (
-                    ex.ErrorCode == Constants.AgentErrorCodes.SessionOwnerMismatch
+                catch (Exception ex) when (
+                    IsSessionOwnerMismatch(ex)
                     && rotateSession != null
                     && rotatedFromSessionId == null)
                 {
-                    // SESSION-OWNER-BINDING: the id on disk belongs to a session this device
-                    // identity does not own. Become a new session and register once more —
-                    // immediately, no backoff, no auth-failure bookkeeping.
+                    // SESSION-OWNER-BINDING: the id on disk names a session this device does not
+                    // own. Become a new session and register once more — immediately, no backoff,
+                    // no auth-failure bookkeeping.
                     rotatedFromSessionId = registration.SessionId;
                     string newSessionId;
                     try
@@ -130,15 +133,25 @@ namespace AutopilotMonitor.Agent.V2.Core.Security
                     catch (Exception rotateEx)
                     {
                         logger.Error($"Session owner mismatch reported by backend but session rotation failed: {rotateEx.Message}", rotateEx);
-                        authFailureTracker?.RecordFailure(ex.StatusCode, "agent/register-session", ex.EndpointUnavailable);
-                        return SessionRegistrationResult.AuthFailed(ex.StatusCode, ex.Message);
+                        if (ex is BackendAuthException authEx)
+                        {
+                            authFailureTracker?.RecordFailure(authEx.StatusCode, "agent/register-session", authEx.EndpointUnavailable);
+                            return SessionRegistrationResult.AuthFailed(authEx.StatusCode, authEx.Message);
+                        }
+                        return SessionRegistrationResult.Failed(ex.Message);
                     }
                     logger.Warning(
                         $"Backend refused registration with {Constants.AgentErrorCodes.SessionOwnerMismatch}: session {rotatedFromSessionId} " +
-                        $"is bound to another device identity (re-enrollment without wipe?). Rotated to {newSessionId}; re-registering.");
+                        $"is not this device's (bound to another device identity after a re-enrollment without wipe, or claimed by another tenant). " +
+                        $"Rotated to {newSessionId}; re-registering.");
                     registration = BuildRegistration(agentConfig, agentVersion, deviceHardware);
                     continue;
                 }
+                catch (BackendRefusalException ex)
+                {
+                    logger.Error($"Session registration refused ({ex.StatusCode}, {ex.ErrorCode}): {ex.Message}");
+                    return SessionRegistrationResult.Failed(ex.Message);
+                }
                 catch (BackendAuthException ex)
                 {
                     logger.Error($"Session registration authentication failed ({ex.StatusCode}): {ex.Message}");
@@ -195,6 +208,12 @@ namespace AutopilotMonitor.Agent.V2.Core.Security
             return SessionRegistrationResult.Failed(lastError ?? "max retries exceeded");
         }
 
+        private static bool IsSessionOwnerMismatch(Exception ex)
+        {
+            var code = (ex as BackendAuthException)?.ErrorCode ?? (ex as BackendRefusalException)?.ErrorCode;
+            return code == Constants.AgentErrorCodes.SessionOwnerMismatch;
+        }
+
         /// <summary>V1 parity exponential backoff: 2^attempt seconds.</summary>
         private static Task DefaultBackoffDelay(int attempt)
             => Task.Delay(((int)Math.Pow(2, attempt)) * 1000);

src/Agent/AutopilotMonitor.Agent.V2.Core/SignalAdapters/EspAndHelloTrackerAdapter.cs

@@ -24,7 +24,7 @@ namespace AutopilotMonitor.Agent.V2.Core.SignalAdapters
     /// <list type="bullet">
     ///   <item><c>HelloCompleted</c> → <see cref="DecisionSignalKind.HelloResolved"/></item>
     ///   <item><c>FinalizingSetupPhaseTriggered</c> → <see cref="DecisionSignalKind.EspPhaseChanged"/> (phase=FinalizingSetup)</item>
-    ///   <item><c>WhiteGloveCompleted</c> → <see cref="DecisionSignalKind.WhiteGloveShellCoreSuccess"/></item>
+    ///   <item><c>WhiteGloveCompleted</c> → <see cref="DecisionSignalKind.WhiteGloveShellCoreSuccess"/> (payload: registry AccountSetup progress at that moment)</item>
     ///   <item><c>EspFailureDetected</c> → <see cref="DecisionSignalKind.EspTerminalFailure"/> (merged aus ShellCore + Provisioning)</item>
     ///   <item><c>DeviceSetupProvisioningComplete</c> → <see cref="DecisionSignalKind.DeviceSetupProvisioningComplete"/></item>
     ///   <item><c>AccountSetupProvisioningComplete</c> → <see cref="DecisionSignalKind.AccountSetupProvisioningComplete"/> (session 330f73f3 fix — strong AccountSetup-done gate)</item>
@@ -46,6 +46,7 @@ namespace AutopilotMonitor.Agent.V2.Core.SignalAdapters
         private readonly EspAndHelloTracker _coordinator;
         private readonly ISignalIngressSink _ingress;
         private readonly IClock _clock;
+        private readonly Func<bool?> _accountSetupProgressProbe;
 
         private bool _helloPosted;
         private bool _helloWizardStartedPosted;
@@ -60,11 +61,14 @@ namespace AutopilotMonitor.Agent.V2.Core.SignalAdapters
         public EspAndHelloTrackerAdapter(
             EspAndHelloTracker coordinator,
             ISignalIngressSink ingress,
-            IClock clock)
+            IClock clock,
+            Func<bool?>? accountSetupProgressProbe = null)
         {
             _coordinator = coordinator ?? throw new ArgumentNullException(nameof(coordinator));
             _ingress = ingress ?? throw new ArgumentNullException(nameof(ingress));
             _clock = clock ?? throw new ArgumentNullException(nameof(clock));
+            // Test seam; production reads the provisioning tracker's registry-derived flag.
+            _accountSetupProgressProbe = accountSetupProgressProbe ?? (() => _coordinator.AccountSetupProgressIfRead);
 
             _coordinator.HelloCompleted += OnHelloCompleted;
             _coordinator.FinalizingSetupPhaseTriggered += OnFinalizing;
@@ -216,13 +220,29 @@ namespace AutopilotMonitor.Agent.V2.Core.SignalAdapters
             if (_whiteGloveSuccessPosted) return;
             _whiteGloveSuccessPosted = true;
 
+            // Registry AccountSetup progress at the moment of the success page. The sealing
+            // classifier lifts its AccountSetup excluder only on an explicit "false" — IME logs
+            // AccountSetup from the device session once the device apps are done, which is no
+            // user activity. Sampled here so replay reads it from the SignalLog (L.2/L.5). An
+            // unread registry sends no key: unknown must keep the excluder, never lift it.
+            var progress = _accountSetupProgressProbe();
+            var progressText = progress.HasValue ? (progress.Value ? "true" : "false") : null;
+
             var now = ResolveOccurredAt(out var derivedFromClock);
             var derivationInputs = new Dictionary<string, string>(StringComparer.Ordinal)
             {
                 ["subSource"] = "ShellCoreTracker",
+                [SignalPayloadKeys.AccountSetupProgress] = progressText ?? "unread",
             };
             TagDerivedTimestamp(derivationInputs, derivedFromClock);
 
+            var payload = progressText == null
+                ? null
+                : new Dictionary<string, string>(StringComparer.Ordinal)
+                {
+                    [SignalPayloadKeys.AccountSetupProgress] = progressText,
+                };
+
             _ingress.Post(
                 kind: DecisionSignalKind.WhiteGloveShellCoreSuccess,
                 occurredAtUtc: now,
@@ -231,7 +251,8 @@ namespace AutopilotMonitor.Agent.V2.Core.SignalAdapters
                     kind: EvidenceKind.Derived,
                     identifier: DetectorId,
                     summary: "WhiteGlove sealing success (coordinator-forwarded)",
-                    derivationInputs: derivationInputs));
+                    derivationInputs: derivationInputs),
+                payload: payload);
         }
 
         private void EmitEspFailure(EspFailureDetectedEventArgs args)

src/Agent/AutopilotMonitor.Agent.V2.Core/Termination/EnrollmentTerminationHandler.cs

@@ -169,6 +169,15 @@ namespace AutopilotMonitor.Agent.V2.Core.Termination
                 _logger.Info(
                     $"EnrollmentTerminationHandler: handling Terminated (reason={args.Reason}, outcome={args.Outcome}, stage={args.StageName}).");
 
+                // The seal is already persisted in the decision snapshot. The marker goes down
+                // before anything else: the technician's reseal can kill this process at any
+                // later step, and a missing marker makes the next boot run the 48-hour break
+                // against the Part-1 session age instead of resuming Part 2.
+                if (isWhiteGlovePart1)
+                {
+                    TrySaveWhiteGloveComplete();
+                }
+
                 var state = TryGetCurrentState();
 
                 // M4.6.δ — shutdown analyzers run BEFORE the dialog / diagnostics upload so
@@ -254,7 +263,8 @@ namespace AutopilotMonitor.Agent.V2.Core.Termination
 
                 // WhiteGlove Part-1 exit: keep the session alive, but announce the handoff so
                 // the timeline clearly marks the transition. The `whiteglove.complete` marker
-                // lets the next agent boot classify itself as a Part-2 resume.
+                // (written at the top of this handler) lets the next agent boot classify itself
+                // as a Part-2 resume.
                 if (isWhiteGlovePart1)
                 {
                     // Stop peripheral collectors BEFORE the Part-1-complete marker so their
@@ -298,7 +308,6 @@ namespace AutopilotMonitor.Agent.V2.Core.Termination
                     DelayLateEventGrace();
                     DrainSpool();
 
-                    TrySaveWhiteGloveComplete();
                     // Option 2 (WG Part 1 graceful-exit hardening, 2026-04-30): write the
                     // clean-exit marker BEFORE _signalShutdown returns control to the main
                     // thread. The AppDomain.ProcessExit handler still writes it as a second

src/Agent/AutopilotMonitor.Agent.V2/Program.Guards.cs

@@ -308,7 +308,7 @@ namespace AutopilotMonitor.Agent.V2
             Func<CleanupService> cleanupServiceFactory,
             AgentLogger logger,
             bool consoleMode,
-            Action onBreakFired = null)
+            Action<double> onBreakFired = null)
         {
             try
             {
@@ -347,7 +347,7 @@ namespace AutopilotMonitor.Agent.V2
                 // (tasks/enrollment-status-reclassification.md). Fired BEFORE cleanup, while the
                 // session state and network are still intact. Never throws — a send failure (e.g. no
                 // network) must not block the cleanup/exit that is the whole point of this guard.
-                try { onBreakFired?.Invoke(); }
+                try { onBreakFired?.Invoke(sessionAgeHours); }
                 catch (Exception cbEx) { logger.Debug($"Emergency-break notify callback failed: {cbEx.Message}"); }
 
                 // Write enrollment-complete.marker so the next start exits cleanly even if

src/Agent/AutopilotMonitor.Agent.V2/Program.cs

@@ -273,7 +273,9 @@ namespace AutopilotMonitor.Agent.V2
             var registration = Runtime.BackendSessionRegistration.Register(
                 bootstrap.AgentConfig, auth, telemetry.MtlsHttpClient, GetAgentVersion(), consoleMode, logger,
                 rotateSession: () => Runtime.BackendSessionRegistration.RotateSession(
-                    bootstrap.AgentConfig, bootstrap.SessionPersistence, auth, transportDir, logger));
+                    bootstrap.AgentConfig, bootstrap.SessionPersistence, auth, transportDir, logger),
+                sessionPersistence: bootstrap.SessionPersistence,
+                configFetchOutcome: runtimeConfig.RemoteConfigService.LastFetchOutcome.ToString());
             if (registration.ShouldExit)
             {
                 return registration.ExitCode;

src/Agent/AutopilotMonitor.Agent.V2/Runtime/AgentBootstrap.cs

@@ -87,7 +87,11 @@ namespace AutopilotMonitor.Agent.V2.Runtime
                     cleanupServiceFactory, logger, consoleMode,
                     // Best-effort: surface the otherwise-silent 48h break to the backend before cleanup
                     // (tasks/enrollment-status-reclassification.md). Never blocks the exit.
-                    onBreakFired: () => EmergencyBreakReporter.TrySend(agentConfig, Program.GetAgentVersion(), logger)))
+                    // The failed-registration record is read here, before the break deletes the
+                    // session files.
+                    onBreakFired: sessionAgeHours => EmergencyBreakReporter.TrySend(
+                        agentConfig, Program.GetAgentVersion(), sessionAgeHours,
+                        sessionPersistence.LoadRegistrationFailure(), logger)))
             {
                 logger.Info("Emergency break fired — agent exiting.");
                 return BootstrapResult.Exit(0);

src/Agent/AutopilotMonitor.Agent.V2/Runtime/AgentRuntimeHost.cs

@@ -526,7 +526,7 @@ namespace AutopilotMonitor.Agent.V2.Runtime
                         // analytics. WhiteGlove Part-2 resume runs as a fresh Classic
                         // enrollment after PR-A's archive-and-reset, so the engine state is
                         // empty and needs the facts re-seeded — same as a first boot.
-                        LifecycleEmitters.PostEnrollmentFactsObserved(orchestrator.IngressSink, logger);
+                        LifecycleEmitters.PostEnrollmentFactsObserved(orchestrator.IngressSink, agentConfig.HelloWaitTimeoutSeconds, logger);
 
                         // V2 parity — post SessionStarted so the reducer establishes the session
                         // anchor (HandleSessionStartedV1 in DecisionEngine.Shared.cs). Skipped only

src/Agent/AutopilotMonitor.Agent.V2/Runtime/BackendSessionRegistration.cs

@@ -1,6 +1,7 @@
 using System;
 using System.Net;
 using System.Net.Http;
+using System.Net.NetworkInformation;
 using System.Threading.Tasks;
 using AutopilotMonitor.Agent.V2.Core.Configuration;
 using AutopilotMonitor.Agent.V2.Core.Logging;
@@ -17,7 +18,8 @@ namespace AutopilotMonitor.Agent.V2.Runtime
     /// <c>IncrementSessionEventCountAsync</c> / <c>UpdateSessionStatusAsync</c> silently
     /// no-op and session status / phase / admin-overrides / validator reconcile all break.
     /// On registration failure we follow V1's rule: collectors MUST NOT start and the
-    /// agent exits cleanly so the next Scheduled-Task tick can retry.
+    /// agent exits cleanly; the Scheduled Task has only a boot trigger, so the next attempt is
+    /// the next boot. The failed run is recorded for the emergency break report.
     /// </summary>
     internal static class BackendSessionRegistration
     {
@@ -28,12 +30,19 @@ namespace AutopilotMonitor.Agent.V2.Runtime
             string agentVersion,
             bool consoleMode,
             AgentLogger logger,
-            Func<string> rotateSession = null)
+            Func<string> rotateSession = null,
+            SessionIdPersistence sessionPersistence = null,
+            string configFetchOutcome = null)
         {
             if (agentConfig == null) throw new ArgumentNullException(nameof(agentConfig));
             if (auth == null) throw new ArgumentNullException(nameof(auth));
             if (logger == null) throw new ArgumentNullException(nameof(logger));
 
+            // Wall clock on purpose: a window far above the retry schedule is the evidence that
+            // the device slept or was suspended while this run tried to register.
+            var startedUtc = DateTime.UtcNow;
+            var linkUpAtStart = TryIsNetworkLinkUp();
+
             var registrationResult = SessionRegistrationHelper.RegisterWithRetryAsync(
                 apiClient: auth.BackendApiClient,
                 agentConfig: agentConfig,
@@ -48,6 +57,10 @@ namespace AutopilotMonitor.Agent.V2.Runtime
                 rotateSession: rotateSession)
                 .GetAwaiter().GetResult();
 
+            RecordRegistrationRun(
+                sessionPersistence, registrationResult, startedUtc, DateTime.UtcNow,
+                linkUpAtStart, TryIsNetworkLinkUp(), configFetchOutcome, logger);
+
             if (registrationResult.Outcome != SessionRegistrationOutcome.Succeeded)
             {
                 logger.Error(
@@ -65,13 +78,71 @@ namespace AutopilotMonitor.Agent.V2.Runtime
             return SessionRegistrationOutcomeResult.Continue(registrationResult);
         }
 
+        /// <summary>Longest <see cref="RegistrationFailureSummary.LastError"/> kept on disk and on the wire.</summary>
+        internal const int MaxRecordedErrorLength = 256;
+
+        /// <summary>
+        /// Keeps <c>registration-failure.json</c> in step with this run: a failed run is folded
+        /// into the record (the emergency break later reports it — for a session that never
+        /// registered, the only account of why), a successful one clears it. Never throws.
+        /// </summary>
+        internal static void RecordRegistrationRun(
+            SessionIdPersistence sessionPersistence,
+            SessionRegistrationResult registrationResult,
+            DateTime startedUtc,
+            DateTime endedUtc,
+            bool? linkUpAtStart,
+            bool? linkUpAtEnd,
+            string configFetchOutcome,
+            AgentLogger logger)
+        {
+            if (sessionPersistence == null || registrationResult == null) return;
+            try
+            {
+                if (registrationResult.Outcome == SessionRegistrationOutcome.Succeeded)
+                {
+                    sessionPersistence.ClearRegistrationFailure(logger);
+                    return;
+                }
+
+                var error = registrationResult.ErrorMessage;
+                if (error != null && error.Length > MaxRecordedErrorLength)
+                    error = error.Substring(0, MaxRecordedErrorLength);
+
+                sessionPersistence.RecordRegistrationFailure(new RegistrationFailureSummary
+                {
+                    FailedRuns = 1,
+                    FirstFailedAtUtc = endedUtc,
+                    LastFailedAtUtc = endedUtc,
+                    Outcome = registrationResult.Outcome.ToString(),
+                    LastError = error,
+                    NetworkLinkUpAtStart = linkUpAtStart,
+                    NetworkLinkUpAtEnd = linkUpAtEnd,
+                    AttemptWindowSeconds = Math.Round(Math.Max(0, (endedUtc - startedUtc).TotalSeconds), 1),
+                    ConfigFetchOutcome = configFetchOutcome,
+                }, logger);
+            }
+            catch (Exception ex)
+            {
+                logger?.Warning($"Registration run record failed (best-effort): {ex.Message}");
+            }
+        }
+
+        private static bool? TryIsNetworkLinkUp()
+        {
+            try { return NetworkInterface.GetIsNetworkAvailable(); }
+            catch { return null; }
+        }
+
         /// <summary>
         /// SESSION-OWNER-BINDING: the one place that turns "backend says this session id is not
         /// ours" into a new session. Rotates the persisted id, points the runtime config at it,
         /// drops the transport spool (its lines carry the refused session's partition key and
         /// would otherwise be uploaded against it after the rotation) and re-targets the
-        /// emergency reporter, which snapshotted the id before registration. Everything else
-        /// reads <c>agentConfig.SessionId</c> live after Phase 6.
+        /// emergency reporter, which snapshotted the id before registration. Everything else that
+        /// holds a session id (orchestrator, emitters, spool) is built in Phase 7 from the rotated
+        /// <c>agentConfig.SessionId</c> — keep it that way: a component built before Phase 6 needs
+        /// its own re-target here.
         /// </summary>
         internal static string RotateSession(
             AgentConfiguration agentConfig,

src/Agent/AutopilotMonitor.Agent.V2/Runtime/EmergencyBreakReporter.cs

@@ -36,15 +36,20 @@ namespace AutopilotMonitor.Agent.V2.Runtime
         // Upper bound for the blocking wait: NIC wait + 3×15s attempts + 2×5s delays + slack.
         private static readonly TimeSpan OverallBudget = TimeSpan.FromSeconds(75);
 
-        public static void TrySend(AgentConfiguration agentConfig, string agentVersion, AgentLogger logger)
+        public static void TrySend(
+            AgentConfiguration agentConfig,
+            string agentVersion,
+            double sessionAgeHours,
+            RegistrationFailureSummary priorRegistrationFailure,
+            AgentLogger logger)
         {
             try
             {
                 WaitForNetwork(logger);
 
                 var auth = BackendClientFactory.BuildAuthClients(agentConfig, agentVersion, logger);
-                var message =
-                    $"Agent absolute session-age emergency break fired (cap {agentConfig.AbsoluteMaxSessionHours}h) — cleaning up and exiting.";
+                var roundedAgeHours = Math.Round(sessionAgeHours, 1);
+                var message = BuildMessage(roundedAgeHours, agentConfig.AbsoluteMaxSessionHours, priorRegistrationFailure);
 
                 // The process is about to exit, so block on the otherwise fire-and-forget send
                 // rather than let it be abandoned mid-flight. TrySendAsync swallows its own HTTP
@@ -56,7 +61,9 @@ namespace AutopilotMonitor.Agent.V2.Runtime
                         message,
                         attempts: SendAttempts,
                         perAttemptTimeout: PerAttemptTimeout,
-                        retryDelay: RetryDelay)
+                        retryDelay: RetryDelay,
+                        sessionAgeHours: roundedAgeHours,
+                        priorRegistrationFailure: priorRegistrationFailure)
                     .Wait(OverallBudget);
             }
             catch (Exception ex)
@@ -65,6 +72,25 @@ namespace AutopilotMonitor.Agent.V2.Runtime
             }
         }
 
+        /// <summary>
+        /// Invariant culture: the text lands in the backend's ops feed and timeline, and a device
+        /// locale would otherwise print "478,2h". The registration clause names what the ops feed
+        /// cannot otherwise show for a session that never registered.
+        /// </summary>
+        internal static string BuildMessage(double roundedAgeHours, int capHours, RegistrationFailureSummary priorRegistrationFailure)
+        {
+            var message = FormattableString.Invariant(
+                $"Agent absolute session-age emergency break fired at session age {roundedAgeHours:F1}h (cap {capHours}h) — cleaning up and exiting.");
+            if (priorRegistrationFailure != null && priorRegistrationFailure.FailedRuns > 0)
+            {
+                message += FormattableString.Invariant(
+                    $" {priorRegistrationFailure.FailedRuns} earlier start(s) failed to register (last: {priorRegistrationFailure.Outcome}, network link at end: {FormatLink(priorRegistrationFailure.NetworkLinkUpAtEnd)}).");
+            }
+            return message;
+        }
+
+        private static string FormatLink(bool? up) => up == null ? "unknown" : (up.Value ? "up" : "down");
+
         /// <summary>
         /// Boot-time NIC grace: the break fires during bootstrap, often seconds after boot while
         /// Wi-Fi is still associating. Polls the cheap link-level signal — it cannot prove backend

src/Agent/AutopilotMonitor.Agent.V2/Runtime/LifecycleEmitters.cs

@@ -314,7 +314,8 @@ namespace AutopilotMonitor.Agent.V2.Runtime
         /// <see cref="DecisionSignalKind.EnrollmentFactsObserved"/> signal carrying the
         /// registry-derived enrollment facts (<c>enrollmentType</c> + <c>isHybridJoin</c>)
         /// so the reducer can seed <see cref="State.EnrollmentScenarioProfile"/> via the
-        /// stage-agnostic <c>HandleEnrollmentFactsObservedV1</c> handler.
+        /// stage-agnostic <c>HandleEnrollmentFactsObservedV1</c> handler — plus the tenant's
+        /// Hello wait when it extends the engine's built-in hello_safety window.
         /// <para>
         /// Posted immediately before <see cref="PostSessionStarted"/> so the Inspector
         /// timeline reads naturally (facts → anchor) — but the reducer correctness does
@@ -329,6 +330,7 @@ namespace AutopilotMonitor.Agent.V2.Runtime
         /// </summary>
         public static void PostEnrollmentFactsObserved(
             ISignalIngressSink ingressSink,
+            int helloWaitTimeoutSeconds,
             AgentLogger logger)
         {
             try
@@ -352,6 +354,16 @@ namespace AutopilotMonitor.Agent.V2.Runtime
                     ["isCloudPc"] = isCloudPc ? "true" : "false",
                 };
 
+                // Tenant Hello wait (remote config, not a registry fact) — stamped only when it
+                // widens the engine's built-in hello_safety window, so sessions on the default
+                // post the same payload as before.
+                var helloWaitSeconds = HelloWaitTimeout.EffectiveSeconds(helloWaitTimeoutSeconds);
+                if (helloWaitSeconds > HelloWaitTimeout.BuiltInSeconds)
+                {
+                    payload[SignalPayloadKeys.HelloWaitTimeoutSeconds] =
+                        helloWaitSeconds.ToString(System.Globalization.CultureInfo.InvariantCulture);
+                }
+
                 var evidence = new Evidence(
                     kind: EvidenceKind.Synthetic,
                     identifier: "enrollment_registry_facts_read",
@@ -364,7 +376,7 @@ namespace AutopilotMonitor.Agent.V2.Runtime
                     evidence: evidence,
                     payload: payload);
 
-                logger.Debug($"EnrollmentFactsObserved signal posted (enrollmentType={enrollmentType}, isHybridJoin={isHybridJoin}, isSelfDeployingProfile={isSelfDeploying}, isCloudPc={isCloudPc}).");
+                logger.Debug($"EnrollmentFactsObserved signal posted (enrollmentType={enrollmentType}, isHybridJoin={isHybridJoin}, isSelfDeployingProfile={isSelfDeploying}, isCloudPc={isCloudPc}, helloWaitSeconds={helloWaitSeconds}).");
             }
             catch (Exception ex)
             {

src/Agent/AutopilotMonitor.DecisionCore.Tests/HelloSafetyTenantWindowTests.cs

@@ -0,0 +1,357 @@
+using System;
+using System.Collections.Generic;
+using System.Globalization;
+using System.Linq;
+using AutopilotMonitor.DecisionCore.Engine;
+using AutopilotMonitor.DecisionCore.Signals;
+using AutopilotMonitor.DecisionCore.State;
+using Xunit;
+
+namespace AutopilotMonitor.DecisionCore.Tests
+{
+    /// <summary>
+    /// Tenant-extendable Hello window. The agent stamps the tenant's <c>HelloWaitTimeoutSeconds</c>
+    /// on <see cref="DecisionSignalKind.EnrollmentFactsObserved"/> when it exceeds the built-in
+    /// 300 s; every <see cref="DeadlineNames.HelloSafety"/> arm site then uses it, clamped to
+    /// 3600 s. Without the key the window stays exactly the built-in 300 s.
+    /// </summary>
+    public sealed class HelloSafetyTenantWindowTests
+    {
+        private static readonly DateTime T0 = new DateTime(2026, 9, 29, 8, 0, 0, DateTimeKind.Utc);
+        private static readonly DateTime EspExit = T0.AddMinutes(6);
+
+        [Theory]
+        [InlineData(null, 300)]
+        [InlineData("1800", 1800)]
+        [InlineData("301", 301)]
+        [InlineData("3600", 3600)]
+        [InlineData("86400", 3600)] // beyond the ceiling — clamped
+        [InlineData("300", 300)]    // not an extension — ignored (the agent never stamps it)
+        [InlineData("120", 300)]
+        public void Esp_exit_arms_hello_safety_with_the_tenant_window(string? helloWait, int expectedSeconds)
+        {
+            var engine = new DecisionEngine();
+            var state = ReduceAll(engine, ClassicUntilEspExit(helloWait));
+
+            Assert.Equal(SessionStage.AwaitingHello, state.Stage);
+            var helloSafety = Assert.Single(state.Deadlines, d => d.Name == DeadlineNames.HelloSafety);
+            Assert.Equal(EspExit.AddSeconds(expectedSeconds), helloSafety.DueAtUtc);
+        }
+
+        [Fact]
+        public void Tenant_window_is_recorded_set_once_with_its_source_ordinal()
+        {
+            var engine = new DecisionEngine();
+            var state = ReduceAll(engine, ClassicUntilEspExit("1800"));
+
+            var fact = state.ScenarioObservations.HelloWaitTimeoutSeconds;
+            Assert.NotNull(fact);
+            Assert.Equal(1800, fact!.Value);
+            Assert.Equal(1, fact.SourceSignalOrdinal);
+
+            // A later re-post (every agent start posts the facts again) never re-bases the window.
+            var repost = engine.Reduce(state, Facts(20, EspExit.AddSeconds(30), "3600"));
+            Assert.Equal(1800, repost.NewState.ScenarioObservations.HelloWaitTimeoutSeconds!.Value);
+            Assert.Equal(1, repost.NewState.ScenarioObservations.HelloWaitTimeoutSeconds!.SourceSignalOrdinal);
+            Assert.Empty(repost.Effects);
+            Assert.Equal(
+                EspExit.AddSeconds(1800),
+                Assert.Single(repost.NewState.Deadlines, d => d.Name == DeadlineNames.HelloSafety).DueAtUtc);
+        }
+
+        [Fact]
+        public void Census_surfaces_the_tenant_window_only_when_recorded()
+        {
+            var engine = new DecisionEngine();
+
+            var extended = DecisionStateSignalCensus.Build(ReduceAll(engine, ClassicUntilEspExit("2700")));
+            Assert.Contains("hello_wait_extended", extended.SignalsSeen);
+            var evidence = Assert.IsAssignableFrom<IDictionary<string, object>>(extended.SignalEvidence["helloWaitExtended"]);
+            Assert.Equal(2700, evidence["seconds"]);
+
+            var builtIn = DecisionStateSignalCensus.Build(ReduceAll(engine, ClassicUntilEspExit(helloWait: null)));
+            Assert.DoesNotContain("hello_wait_extended", builtIn.SignalsSeen);
+        }
+
+        [Theory]
+        [InlineData("abc")]
+        [InlineData("")]
+        [InlineData("0")]
+        [InlineData("-5")]
+        public void Malformed_tenant_window_is_ignored(string helloWait)
+        {
+            var engine = new DecisionEngine();
+            var state = ReduceAll(engine, ClassicUntilEspExit(helloWait));
+
+            Assert.Null(state.ScenarioObservations.HelloWaitTimeoutSeconds);
+            Assert.Equal(
+                EspExit.AddSeconds(300),
+                Assert.Single(state.Deadlines, d => d.Name == DeadlineNames.HelloSafety).DueAtUtc);
+        }
+
+        [Fact]
+        public void Late_tenant_window_extends_the_armed_hello_safety_from_its_original_anchor()
+        {
+            // A backfilled ESP exit armed the built-in window before the facts signal landed.
+            var engine = new DecisionEngine();
+            var state = ReduceAll(engine, ClassicUntilEspExit(helloWait: null));
+            Assert.Equal(
+                EspExit.AddSeconds(300),
+                Assert.Single(state.Deadlines, d => d.Name == DeadlineNames.HelloSafety).DueAtUtc);
+
+            var step = engine.Reduce(state, Facts(20, EspExit.AddSeconds(10), "1800"));
+
+            Assert.Equal(SessionStage.AwaitingHello, step.NewState.Stage);
+            var extended = Assert.Single(step.NewState.Deadlines, d => d.Name == DeadlineNames.HelloSafety);
+            Assert.Equal(EspExit.AddSeconds(1800), extended.DueAtUtc);
+            Assert.Equal(DecisionSignalKind.DeadlineFired, extended.FiresSignalKind);
+            Assert.Equal(DeadlineNames.HelloSafety, extended.FiresPayload![SignalPayloadKeys.Deadline]);
+
+            // The live scheduler is told to replace its timer (same name → replaced).
+            var effect = Assert.Single(step.Effects);
+            Assert.Equal(DecisionEffectKind.ScheduleDeadline, effect.Kind);
+            Assert.Equal(EspExit.AddSeconds(1800), effect.Deadline!.DueAtUtc);
+        }
+
+        [Theory]
+        [InlineData("300")]
+        [InlineData("120")]
+        public void Late_tenant_window_not_longer_than_the_built_in_leaves_the_deadline_alone(string helloWait)
+        {
+            var engine = new DecisionEngine();
+            var state = ReduceAll(engine, ClassicUntilEspExit(helloWait: null));
+
+            var step = engine.Reduce(state, Facts(20, EspExit.AddSeconds(10), helloWait));
+
+            Assert.Empty(step.Effects);
+            Assert.Null(step.NewState.ScenarioObservations.HelloWaitTimeoutSeconds);
+            Assert.Equal(
+                EspExit.AddSeconds(300),
+                Assert.Single(step.NewState.Deadlines, d => d.Name == DeadlineNames.HelloSafety).DueAtUtc);
+        }
+
+        [Fact]
+        public void Fire_of_the_replaced_built_in_deadline_is_dead_ended_after_the_late_extension()
+        {
+            // The old timer fired while the facts signal was still queued ahead of it: the facts
+            // step moves hello_safety out, then the queued fire of the replaced incarnation
+            // arrives. It must not time Hello out at the built-in window.
+            var engine = new DecisionEngine();
+            var state = ReduceAll(engine, ClassicUntilEspExit(helloWait: null));
+            var builtInDue = EspExit.AddSeconds(300);
+            state = engine.Reduce(state, Facts(20, EspExit.AddSeconds(10), "1800")).NewState;
+
+            var stale = engine.Reduce(state, MakeSignal(21, DecisionSignalKind.DeadlineFired, builtInDue.AddSeconds(2),
+                Fired(DeadlineNames.HelloSafety, builtInDue)));
+
+            Assert.False(stale.Transition.Taken);
+            Assert.Equal("hello_safety_stale_superseded_by_rearm", stale.Transition.DeadEndReason);
+            Assert.Empty(stale.Effects);
+            Assert.Null(stale.NewState.HelloResolvedUtc);
+            Assert.Equal(SessionStage.AwaitingHello, stale.NewState.Stage);
+            var extendedDue = EspExit.AddSeconds(1800);
+            Assert.Equal(
+                extendedDue,
+                Assert.Single(stale.NewState.Deadlines, d => d.Name == DeadlineNames.HelloSafety).DueAtUtc);
+
+            // The fire of the current incarnation still times Hello out (desktop is in → Finalizing).
+            var current = engine.Reduce(stale.NewState, MakeSignal(22, DecisionSignalKind.DeadlineFired, extendedDue,
+                Fired(DeadlineNames.HelloSafety, extendedDue)));
+            Assert.Equal(SessionStage.Finalizing, current.NewState.Stage);
+            Assert.Equal("Timeout", current.NewState.HelloOutcome!.Value);
+        }
+
+        [Fact]
+        public void Late_tenant_window_after_the_built_in_expired_leaves_the_pending_fire_to_decide()
+        {
+            var engine = new DecisionEngine();
+            var state = ReduceAll(engine, ClassicUntilEspExit(helloWait: null));
+
+            var step = engine.Reduce(state, Facts(20, EspExit.AddSeconds(301), "1800"));
+
+            Assert.Empty(step.Effects);
+            Assert.Equal(
+                EspExit.AddSeconds(300),
+                Assert.Single(step.NewState.Deadlines, d => d.Name == DeadlineNames.HelloSafety).DueAtUtc);
+            // The fact is still recorded — later arm sites use it.
+            Assert.Equal(1800, step.NewState.ScenarioObservations.HelloWaitTimeoutSeconds!.Value);
+        }
+
+        [Fact]
+        public void Late_extension_is_replay_deterministic()
+        {
+            var engine = new DecisionEngine();
+            var signals = ClassicUntilEspExit(helloWait: null)
+                .Concat(new[] { Facts(20, EspExit.AddSeconds(10), "2400") })
+                .ToList();
+
+            var stepwise = ReduceAll(engine, signals);
+            var replayed = ReducerReplay.Replay(engine, Seed(), signals);
+
+            Assert.Equal(
+                Assert.Single(stepwise.Deadlines, d => d.Name == DeadlineNames.HelloSafety).DueAtUtc,
+                Assert.Single(replayed.Deadlines, d => d.Name == DeadlineNames.HelloSafety).DueAtUtc);
+            Assert.Equal(EspExit.AddSeconds(2400), replayed.Deadlines.Single(d => d.Name == DeadlineNames.HelloSafety).DueAtUtc);
+        }
+
+        [Fact]
+        public void Tenant_window_also_applies_to_the_AccountSetup_complete_deferred_promote()
+        {
+            // Strong gate arrives after the final exit and before the desktop: the deferred
+            // promote arms hello_safety at the gate signal — with the tenant window.
+            var engine = new DecisionEngine();
+            var state = ReduceAll(engine, new[]
+            {
+                MakeSignal(0, DecisionSignalKind.SessionStarted, T0),
+                Facts(1, T0.AddSeconds(5), "2400"),
+                MakeSignal(2, DecisionSignalKind.EspPhaseChanged, T0.AddMinutes(1), Phase("DeviceSetup")),
+                MakeSignal(3, DecisionSignalKind.HelloPolicyDetected, T0.AddMinutes(1).AddSeconds(5),
+                    new Dictionary<string, string> { [SignalPayloadKeys.HelloEnabled] = "true" }),
+                MakeSignal(4, DecisionSignalKind.EspPhaseChanged, T0.AddMinutes(12), Phase("AccountSetup")),
+                MakeSignal(5, DecisionSignalKind.EspExiting, T0.AddMinutes(24)),
+            });
+            Assert.Equal(SessionStage.EspAccountSetup, state.Stage);
+
+            var gate = T0.AddMinutes(25);
+            var step = engine.Reduce(state, MakeSignal(6, DecisionSignalKind.AccountSetupProvisioningComplete, gate));
+
+            Assert.Equal(SessionStage.AwaitingHello, step.NewState.Stage);
+            Assert.Equal(
+                gate.AddSeconds(2400),
+                Assert.Single(step.NewState.Deadlines, d => d.Name == DeadlineNames.HelloSafety).DueAtUtc);
+        }
+
+        [Theory]
+        [InlineData(null, 30 * 60)]
+        [InlineData("1200", 30 * 60)] // a shorter tenant window never shortens the WDP backstop
+        [InlineData("3600", 3600)]
+        public void Device_preparation_backstop_is_never_shorter_than_the_tenant_window(string? helloWait, int expectedSeconds)
+        {
+            // WDP has no ESP, so hello_safety never arms; the desktop-armed backstop is the last
+            // net behind the tracker's Hello completion timer, which runs on the tenant budget.
+            var engine = new DecisionEngine();
+            var facts = new Dictionary<string, string>
+            {
+                [SignalPayloadKeys.EnrollmentType] = "v2",
+                [SignalPayloadKeys.EnrollmentTypeDeterministic] = "true",
+            };
+            if (helloWait != null)
+                facts[SignalPayloadKeys.HelloWaitTimeoutSeconds] = helloWait;
+            var desktop = T0.AddMinutes(9);
+
+            var state = ReduceAll(engine, new[]
+            {
+                MakeSignal(0, DecisionSignalKind.SessionStarted, T0),
+                MakeSignal(1, DecisionSignalKind.EnrollmentFactsObserved, T0, facts),
+                MakeSignal(2, DecisionSignalKind.DesktopArrived, desktop),
+            });
+
+            Assert.Equal(
+                desktop.AddSeconds(expectedSeconds),
+                Assert.Single(state.Deadlines, d => d.Name == DeadlineNames.DevicePrepCompletion).DueAtUtc);
+        }
+
+        [Fact]
+        public void IsTenantHelloWaitPending_is_true_only_while_an_extended_window_is_armed_and_not_yet_due()
+        {
+            var engine = new DecisionEngine();
+            var awaiting = ReduceAll(engine, ClassicUntilEspExit("1800"));
+            var due = EspExit.AddSeconds(1800);
+
+            Assert.True(DecisionEngine.IsTenantHelloWaitPending(awaiting, EspExit.AddMinutes(1)));
+            Assert.True(DecisionEngine.IsTenantHelloWaitPending(awaiting, due.AddSeconds(-1)));
+            Assert.False(DecisionEngine.IsTenantHelloWaitPending(awaiting, due));
+            Assert.False(DecisionEngine.IsTenantHelloWaitPending(awaiting, due.AddSeconds(1)));
+            Assert.False(DecisionEngine.IsTenantHelloWaitPending(null, EspExit));
+            Assert.False(DecisionEngine.IsTenantHelloWaitPending(Seed(), EspExit));
+
+            // The built-in window never counts: default tenants keep their stall reporting.
+            var builtIn = ReduceAll(engine, ClassicUntilEspExit(helloWait: null));
+            Assert.Contains(builtIn.Deadlines, d => d.Name == DeadlineNames.HelloSafety);
+            Assert.False(DecisionEngine.IsTenantHelloWaitPending(builtIn, EspExit.AddMinutes(1)));
+
+            // Hello resolved → the deadline is cancelled → nothing pending any more.
+            var resolved = engine.Reduce(awaiting, MakeSignal(20, DecisionSignalKind.HelloResolved, EspExit.AddMinutes(10),
+                new Dictionary<string, string> { [SignalPayloadKeys.HelloOutcome] = "completed" })).NewState;
+            Assert.False(DecisionEngine.IsTenantHelloWaitPending(resolved, EspExit.AddMinutes(11)));
+        }
+
+        [Fact]
+        public void IsTenantHelloWaitPending_covers_the_device_preparation_backstop()
+        {
+            var engine = new DecisionEngine();
+            var desktop = T0.AddMinutes(9);
+            var state = ReduceAll(engine, new[]
+            {
+                MakeSignal(0, DecisionSignalKind.SessionStarted, T0),
+                MakeSignal(1, DecisionSignalKind.EnrollmentFactsObserved, T0, new Dictionary<string, string>
+                {
+                    [SignalPayloadKeys.EnrollmentType] = "v2",
+                    [SignalPayloadKeys.EnrollmentTypeDeterministic] = "true",
+                    [SignalPayloadKeys.HelloWaitTimeoutSeconds] = "3600",
+                }),
+                MakeSignal(2, DecisionSignalKind.DesktopArrived, desktop),
+            });
+
+            Assert.True(DecisionEngine.IsTenantHelloWaitPending(state, desktop.AddMinutes(59)));
+            Assert.False(DecisionEngine.IsTenantHelloWaitPending(state, desktop.AddMinutes(60)));
+        }
+
+        // ============================================================ helpers
+
+        private static DecisionState Seed() => DecisionState.CreateInitial("sess-hello-window", "tenant", T0);
+
+        /// <summary>
+        /// Classic user-driven flow up to the final ESP exit: strong gate and desktop in place,
+        /// Hello policy unknown — so the exit promotes to AwaitingHello and arms hello_safety.
+        /// </summary>
+        private static List<DecisionSignal> ClassicUntilEspExit(string? helloWait)
+        {
+            var signals = new List<DecisionSignal> { MakeSignal(0, DecisionSignalKind.SessionStarted, T0) };
+            if (helloWait != null)
+                signals.Add(Facts(1, T0.AddSeconds(5), helloWait));
+            signals.Add(MakeSignal(2, DecisionSignalKind.EspPhaseChanged, T0.AddMinutes(1), Phase("AccountSetup")));
+            signals.Add(MakeSignal(3, DecisionSignalKind.AccountSetupProvisioningComplete, T0.AddMinutes(2)));
+            signals.Add(MakeSignal(4, DecisionSignalKind.DesktopArrived, T0.AddMinutes(3)));
+            signals.Add(MakeSignal(5, DecisionSignalKind.EspExiting, EspExit));
+            return signals;
+        }
+
+        private static DecisionState ReduceAll(DecisionEngine engine, IEnumerable<DecisionSignal> signals)
+        {
+            var state = Seed();
+            foreach (var signal in signals)
+                state = engine.Reduce(state, signal).NewState;
+            return state;
+        }
+
+        private static DecisionSignal Facts(long ordinal, DateTime occurredAtUtc, string helloWait) =>
+            MakeSignal(ordinal, DecisionSignalKind.EnrollmentFactsObserved, occurredAtUtc,
+                new Dictionary<string, string> { [SignalPayloadKeys.HelloWaitTimeoutSeconds] = helloWait });
+
+        private static Dictionary<string, string> Fired(string deadlineName, DateTime dueAtUtc) =>
+            new Dictionary<string, string>
+            {
+                [SignalPayloadKeys.Deadline] = deadlineName,
+                [SignalPayloadKeys.DeadlineDueAtUtc] = dueAtUtc.ToString("O", CultureInfo.InvariantCulture),
+            };
+
+        private static Dictionary<string, string> Phase(string phase) =>
+            new Dictionary<string, string> { [SignalPayloadKeys.EspPhase] = phase };
+
+        private static DecisionSignal MakeSignal(
+            long ordinal,
+            DecisionSignalKind kind,
+            DateTime occurredAtUtc,
+            IReadOnlyDictionary<string, string>? payload = null) =>
+            new DecisionSignal(
+                sessionSignalOrdinal: ordinal,
+                sessionTraceOrdinal: ordinal,
+                kind: kind,
+                kindSchemaVersion: 1,
+                occurredAtUtc: occurredAtUtc,
+                sourceOrigin: "test",
+                evidence: new Evidence(EvidenceKind.Synthetic, $"t-{kind}-{ordinal}", $"synthetic {kind}"),
+                payload: payload);
+    }
+}

src/Agent/AutopilotMonitor.DecisionCore.Tests/SerializationTests.cs

@@ -522,7 +522,9 @@ namespace AutopilotMonitor.DecisionCore.Tests
                     .WithSkipUserEsp(value: true, sourceSignalOrdinal: 4)
                     .WithEspSyncFailureTimeoutMinutes(value: 90, sourceSignalOrdinal: 5)
                     .WithRegistrySelfDeployingProfile(value: false, sourceSignalOrdinal: 6)
-                    .WithCloudPc(value: true, sourceSignalOrdinal: 7))
+                    .WithCloudPc(value: true, sourceSignalOrdinal: 7)
+                    .WithAccountSetupProgressAtWhiteGloveSuccess(value: false, sourceSignalOrdinal: 8)
+                    .WithHelloWaitTimeoutSeconds(value: 1800, sourceSignalOrdinal: 9))
                 .Build();
 
             var json = StateSerializer.Serialize(state);
@@ -538,6 +540,10 @@ namespace AutopilotMonitor.DecisionCore.Tests
             Assert.Equal(6, o.RegistrySelfDeployingProfile!.SourceSignalOrdinal);
             Assert.True(o.CloudPc!.Value);
             Assert.Equal(7, o.CloudPc!.SourceSignalOrdinal);
+            Assert.False(o.AccountSetupProgressAtWhiteGloveSuccess!.Value);
+            Assert.Equal(8, o.AccountSetupProgressAtWhiteGloveSuccess!.SourceSignalOrdinal);
+            Assert.Equal(1800, o.HelloWaitTimeoutSeconds!.Value);
+            Assert.Equal(9, o.HelloWaitTimeoutSeconds!.SourceSignalOrdinal);
 
             // Never-observed facts must rehydrate as null (null ≠ veto — session 62e603c9).
             Assert.Null(o.WhiteGloveSealingPatternSeen);

src/Agent/AutopilotMonitor.DecisionCore.Tests/SignalKindFixtureCoverageTests.cs

@@ -31,7 +31,6 @@ namespace AutopilotMonitor.DecisionCore.Tests
             DecisionSignalKind.HelloResolved,
             DecisionSignalKind.ImeUserSessionCompleted,
             DecisionSignalKind.DeviceSetupProvisioningComplete,
-            DecisionSignalKind.WhiteGloveShellCoreSuccess,
             DecisionSignalKind.WhiteGloveSealingPatternDetected,
             DecisionSignalKind.AadUserJoinedLate,
             DecisionSignalKind.DeviceInfoCollected,

src/Agent/AutopilotMonitor.DecisionCore.Tests/WhiteGloveAccountSetupProgressGateTests.cs

@@ -0,0 +1,155 @@
+using System;
+using System.Collections.Generic;
+using AutopilotMonitor.DecisionCore.Classifiers;
+using AutopilotMonitor.DecisionCore.Engine;
+using AutopilotMonitor.DecisionCore.Serialization;
+using AutopilotMonitor.DecisionCore.Signals;
+using AutopilotMonitor.DecisionCore.State;
+using Xunit;
+
+namespace AutopilotMonitor.DecisionCore.Tests
+{
+    /// <summary>
+    /// The sealing classifier's AccountSetup excluder (-40) applies only when the IME reported
+    /// the AccountSetup phase AND the WhiteGlove_Success signal did not report an untouched
+    /// AccountSetup registry. IME logs "EspPhase: AccountSetup" from the device session once
+    /// the device apps are done — on pre-provisioning with the user ESP enabled that line
+    /// routinely precedes the technician's success page and used to leave Part 1 unsealed.
+    /// The gate may only ever lift the excluder, never add one.
+    /// </summary>
+    public sealed class WhiteGloveAccountSetupProgressGateTests
+    {
+        private static readonly DateTime T0 = new DateTime(2026, 9, 29, 7, 25, 9, DateTimeKind.Utc);
+
+        [Fact]
+        public void Ime_account_setup_with_untouched_registry_seals_on_the_fast_path()
+        {
+            var step = new DecisionEngine().Reduce(
+                StateWithImeAccountSetup(), MakeShellCoreSignal(ordinal: 3, accountSetupProgress: "false"));
+
+            Assert.Equal(SessionStage.WhiteGloveSealed, step.NewState.Stage);
+            Assert.Equal(SessionOutcome.WhiteGlovePart1Sealed, step.NewState.Outcome);
+            Assert.Equal(HypothesisLevel.Confirmed, step.NewState.ClassifierOutcomes.WhiteGloveSealing.Level);
+            Assert.Equal(WhiteGloveSealingClassifier.WeightShellCoreWhiteGloveSuccess,
+                step.NewState.ClassifierOutcomes.WhiteGloveSealing.Score);
+
+            var fact = step.NewState.ScenarioObservations.AccountSetupProgressAtWhiteGloveSuccess;
+            Assert.NotNull(fact);
+            Assert.False(fact!.Value);
+            Assert.Equal(3, fact.SourceSignalOrdinal);
+        }
+
+        [Fact]
+        public void Ime_account_setup_with_registry_progress_keeps_the_excluder()
+        {
+            var step = new DecisionEngine().Reduce(
+                StateWithImeAccountSetup(), MakeShellCoreSignal(ordinal: 3, accountSetupProgress: "true"));
+
+            Assert.Equal(SessionStage.EspAccountSetup, step.NewState.Stage);
+            Assert.Contains(step.Effects, e => e.Kind == DecisionEffectKind.RunClassifier);
+            Assert.True(DecisionEngine.BuildWhiteGloveSealingSnapshot(step.NewState).HasAccountSetupActivity);
+        }
+
+        [Fact]
+        public void Missing_progress_payload_keeps_the_excluder_for_older_agents()
+        {
+            var step = new DecisionEngine().Reduce(
+                StateWithImeAccountSetup(), MakeShellCoreSignal(ordinal: 3, accountSetupProgress: null));
+
+            Assert.Equal(SessionStage.EspAccountSetup, step.NewState.Stage);
+            Assert.Null(step.NewState.ScenarioObservations.AccountSetupProgressAtWhiteGloveSuccess);
+            Assert.True(DecisionEngine.BuildWhiteGloveSealingSnapshot(step.NewState).HasAccountSetupActivity);
+        }
+
+        [Theory]
+        [InlineData("")]
+        [InlineData("yes")]
+        public void Unparseable_progress_payload_is_treated_as_missing(string raw)
+        {
+            var step = new DecisionEngine().Reduce(
+                StateWithImeAccountSetup(), MakeShellCoreSignal(ordinal: 3, accountSetupProgress: raw));
+
+            Assert.Null(step.NewState.ScenarioObservations.AccountSetupProgressAtWhiteGloveSuccess);
+            Assert.NotEqual(SessionStage.WhiteGloveSealed, step.NewState.Stage);
+        }
+
+        [Fact]
+        public void Registry_progress_without_ime_account_setup_adds_no_excluder()
+        {
+            var state = DecisionState.CreateInitial("s-gate", "t-gate").ToBuilder()
+                .WithStage(SessionStage.EspDeviceSetup)
+                .WithStepIndex(1)
+                .WithLastAppliedSignalOrdinal(0)
+                .Build();
+
+            var step = new DecisionEngine().Reduce(state, MakeShellCoreSignal(ordinal: 1, accountSetupProgress: "true"));
+
+            Assert.Equal(SessionStage.WhiteGloveSealed, step.NewState.Stage);
+        }
+
+        [Fact]
+        public void Untouched_registry_does_not_lift_a_hard_excluder()
+        {
+            var builder = DecisionState.CreateInitial("s-gate", "t-gate").ToBuilder()
+                .WithStage(SessionStage.EspAccountSetup)
+                .WithStepIndex(3)
+                .WithLastAppliedSignalOrdinal(2);
+            builder.AccountSetupEnteredUtc = new SignalFact<DateTime>(T0.AddMinutes(-3), 1);
+            builder.DesktopArrivedUtc = new SignalFact<DateTime>(T0.AddMinutes(-1), 2);
+
+            var step = new DecisionEngine().Reduce(builder.Build(), MakeShellCoreSignal(ordinal: 3, accountSetupProgress: "false"));
+
+            Assert.NotEqual(SessionStage.WhiteGloveSealed, step.NewState.Stage);
+            var snapshot = DecisionEngine.BuildWhiteGloveSealingSnapshot(step.NewState);
+            Assert.False(snapshot.HasAccountSetupActivity);
+            Assert.True(snapshot.DesktopArrived);
+        }
+
+        [Fact]
+        public void Progress_fact_survives_the_state_serializer()
+        {
+            var step = new DecisionEngine().Reduce(
+                StateWithImeAccountSetup(), MakeShellCoreSignal(ordinal: 3, accountSetupProgress: "false"));
+
+            var roundTripped = StateSerializer.Deserialize(StateSerializer.Serialize(step.NewState));
+
+            var fact = roundTripped.ScenarioObservations.AccountSetupProgressAtWhiteGloveSuccess;
+            Assert.NotNull(fact);
+            Assert.False(fact!.Value);
+            Assert.Equal(3, fact.SourceSignalOrdinal);
+        }
+
+        private static DecisionState StateWithImeAccountSetup()
+        {
+            // The IME device-session line: AccountSetup entered before the success page.
+            var builder = DecisionState.CreateInitial("s-gate", "t-gate").ToBuilder()
+                .WithStage(SessionStage.EspAccountSetup)
+                .WithStepIndex(3)
+                .WithLastAppliedSignalOrdinal(2);
+            builder.AccountSetupEnteredUtc = new SignalFact<DateTime>(T0.AddMinutes(-3), 2);
+            return builder.Build();
+        }
+
+        private static DecisionSignal MakeShellCoreSignal(long ordinal, string? accountSetupProgress)
+        {
+            var payload = new Dictionary<string, string>(StringComparer.Ordinal);
+            if (accountSetupProgress != null)
+            {
+                payload[SignalPayloadKeys.AccountSetupProgress] = accountSetupProgress;
+            }
+
+            return new DecisionSignal(
+                sessionSignalOrdinal: ordinal,
+                sessionTraceOrdinal: ordinal,
+                kind: DecisionSignalKind.WhiteGloveShellCoreSuccess,
+                kindSchemaVersion: 1,
+                occurredAtUtc: T0,
+                sourceOrigin: "EspAndHelloTracker",
+                evidence: new Evidence(
+                    kind: EvidenceKind.Raw,
+                    identifier: "ShellCore-62407",
+                    summary: "WhiteGlove_Success"),
+                payload: payload);
+        }
+    }
+}

src/Backend/AutopilotMonitor.Functions.Tests/AdminConfigurationRateLimitSyncTests.cs

@@ -57,8 +57,6 @@ public class AdminConfigurationRateLimitSyncTests
             It.IsAny<Func<AdminConfiguration, string?>>(), "global-admin@contoso.com", It.IsAny<string?>()), Times.Once);
         // ...but NO tenant configuration is enumerated or mutated (sync removed).
         repo.Verify(r => r.GetAllTenantConfigurationsAsync(), Times.Never);
-        repo.Verify(r => r.SaveTenantConfigurationAsync(It.IsAny<TenantConfiguration>()), Times.Never);
-        repo.Verify(r => r.SaveTenantConfigurationAsync(
-            It.IsAny<TenantConfiguration>(), It.IsAny<string?>(), It.IsAny<string?>()), Times.Never);
+        repo.VerifyNoTenantConfigWrite();
     }
 }

src/Backend/AutopilotMonitor.Functions.Tests/AppHomingServiceTests.cs

@@ -1,4 +1,5 @@
 using AutopilotMonitor.Functions.Security;
+using AutopilotMonitor.Functions.Functions.Admin;
 using AutopilotMonitor.Functions.Services;
 using AutopilotMonitor.Functions.Services.GraphResolution;
 using AutopilotMonitor.Functions.Services.Notifications;
@@ -32,6 +33,7 @@ public class AppHomingServiceTests
 
     private readonly Mock<AdminConfigurationService> _adminConfigMock;
     private readonly Mock<TenantConfigurationService> _tenantConfigMock;
+    private readonly StoredTenantConfig _store;
     private readonly Mock<GraphTokenService> _graphTokenMock;
     private readonly Mock<IGraphFeatureDetector> _detectorMock;
     private readonly Mock<IMaintenanceRepository> _maintenanceMock;
@@ -99,13 +101,10 @@ public class AppHomingServiceTests
             new TelemetryClient(new TelemetryConfiguration { DisableTelemetry = true }));
 
         // Defaults: flag on, legacy-homed config exists, both apps mint a role-less token (the
-        // superset rule passes trivially). FlipAsync reads via the cache-bypassing
-        // GetConfigurationFreshAsync (read-modify-write on the whole entity).
+        // superset rule passes trivially). FlipAsync writes through UpdateAsync (conditional
+        // write on the fresh row), stubbed over one stored row.
         _adminConfigMock.Setup(x => x.IsSelfServiceAppHomingEnabledAsync()).ReturnsAsync(true);
-        var config = LegacyHomedConfig();
-        _tenantConfigMock.Setup(x => x.GetConfigurationFreshAsync(TenantId)).ReturnsAsync(config);
-        _tenantConfigMock.Setup(x => x.SaveConfigurationAsync(It.IsAny<TenantConfiguration>(), It.IsAny<string?>(), It.IsAny<string?>()))
-            .Returns(Task.CompletedTask);
+        _store = _tenantConfigMock.StubWrites(TenantId, LegacyHomedConfig());
         SetupProbe(GraphTokenResult.Success(Jwt()));
         SetupLegacyToken(GraphTokenResult.Success(Jwt()));
     }
@@ -143,8 +142,7 @@ public class AppHomingServiceTests
                 It.IsAny<CancellationToken>()))
             .ReturnsAsync(result);
 
-    private void VerifyNeverSaved() =>
-        _tenantConfigMock.Verify(x => x.SaveConfigurationAsync(It.IsAny<TenantConfiguration>(), It.IsAny<string?>(), It.IsAny<string?>()), Times.Never);
+    private void VerifyNeverSaved() => Assert.Empty(_store.Writes);
 
     // ── IsFunnelEligibleAsync ───────────────────────────────────────────────
 
@@ -340,9 +338,10 @@ public class AppHomingServiceTests
         var outcome = await _sut.TryAutoFlipToPrimaryAsync(LegacyHomedConfig(), Actor);
 
         Assert.Equal(AppHomingAutoFlipOutcome.Flipped, outcome.Outcome);
-        _tenantConfigMock.Verify(x => x.SaveConfigurationAsync(It.Is<TenantConfiguration>(c =>
-            string.Equals(c.HomedAppClientId, PrimaryId, StringComparison.OrdinalIgnoreCase)
-            && c.UpdatedBy == Actor), It.IsAny<string?>(), It.IsAny<string?>()), Times.Once);
+        var (written, source, _) = Assert.Single(_store.Writes);
+        Assert.Equal(PrimaryId, written.HomedAppClientId, ignoreCase: true);
+        Assert.Equal(Actor, written.UpdatedBy);
+        Assert.Equal("app-homing", source);
         _detectorMock.Verify(x => x.InvalidateTenant(TenantId), Times.Once);
         _maintenanceMock.Verify(m => m.LogAuditEntryAsync(TenantId, "UPDATE", "TenantConfiguration",
             TenantId, Actor, It.Is<Dictionary<string, string>?>(d =>
@@ -359,7 +358,7 @@ public class AppHomingServiceTests
         var outcome = await _sut.TryAutoFlipToPrimaryAsync(LegacyHomedConfig(), Actor);
 
         Assert.Equal(AppHomingAutoFlipOutcome.ProbeFailed, outcome.Outcome);
-        _tenantConfigMock.Verify(x => x.SaveConfigurationAsync(It.IsAny<TenantConfiguration>(), It.IsAny<string?>(), It.IsAny<string?>()), Times.Never);
+        VerifyNeverSaved();
     }
 
     [Fact]
@@ -370,7 +369,7 @@ public class AppHomingServiceTests
         var outcome = await _sut.TryAutoFlipToPrimaryAsync(LegacyHomedConfig(), Actor);
 
         Assert.Equal(AppHomingAutoFlipOutcome.ProbeTransient, outcome.Outcome);
-        _tenantConfigMock.Verify(x => x.SaveConfigurationAsync(It.IsAny<TenantConfiguration>(), It.IsAny<string?>(), It.IsAny<string?>()), Times.Never);
+        VerifyNeverSaved();
     }
 
     [Fact]
@@ -394,7 +393,7 @@ public class AppHomingServiceTests
         var outcome = await _sut.TryAutoFlipToPrimaryAsync(LegacyHomedConfig(), Actor);
 
         Assert.Equal(AppHomingAutoFlipOutcome.NotEligible, outcome.Outcome);
-        _tenantConfigMock.Verify(x => x.SaveConfigurationAsync(It.IsAny<TenantConfiguration>(), It.IsAny<string?>(), It.IsAny<string?>()), Times.Never);
+        VerifyNeverSaved();
     }
 
     // ── FlipAsync ───────────────────────────────────────────────────────────
@@ -404,7 +403,7 @@ public class AppHomingServiceTests
     {
         var config = LegacyHomedConfig();
         config.EntraAppRolesEnabled = true;
-        _tenantConfigMock.Setup(x => x.GetConfigurationFreshAsync(TenantId)).ReturnsAsync(config);
+        _store.Row = config;
 
         await _sut.FlipAsync(TenantId, PrimaryId, Actor, "manual-ga");
 
@@ -417,11 +416,11 @@ public class AppHomingServiceTests
     {
         var config = LegacyHomedConfig();
         config.HomedAppClientId = PrimaryId;
-        _tenantConfigMock.Setup(x => x.GetConfigurationFreshAsync(TenantId)).ReturnsAsync(config);
+        _store.Row = config;
 
         await _sut.FlipAsync(TenantId, PrimaryId, Actor, "manual-ga");
 
-        _tenantConfigMock.Verify(x => x.SaveConfigurationAsync(It.IsAny<TenantConfiguration>(), It.IsAny<string?>(), It.IsAny<string?>()), Times.Never);
+        VerifyNeverSaved();
         _detectorMock.Verify(x => x.InvalidateTenant(It.IsAny<string>()), Times.Never);
         Assert.Empty(_savedOpsEvents);
     }
@@ -430,30 +429,68 @@ public class AppHomingServiceTests
     public async Task Flip_throws_when_config_row_missing_and_saves_nothing()
     {
         // A flip must never materialize a default row (e.g. resurrect an offboarded tenant);
-        // the fresh read returning null is a hard stop.
-        _tenantConfigMock.Setup(x => x.GetConfigurationFreshAsync(TenantId))
-            .ReturnsAsync((TenantConfiguration?)null);
+        // a missing row is a hard stop.
+        _store.Row = null;
 
         await Assert.ThrowsAsync<InvalidOperationException>(
             () => _sut.FlipAsync(TenantId, PrimaryId, Actor, "manual-ga"));
 
-        _tenantConfigMock.Verify(x => x.SaveConfigurationAsync(It.IsAny<TenantConfiguration>(), It.IsAny<string?>(), It.IsAny<string?>()), Times.Never);
+        VerifyNeverSaved();
         _detectorMock.Verify(x => x.InvalidateTenant(It.IsAny<string>()), Times.Never);
         Assert.Empty(_savedOpsEvents);
     }
 
+    [Fact]
+    public async Task Flip_decides_on_the_stored_row_not_the_callers_view()
+    {
+        // The caller's (cached) view is legacy-homed; the stored row was flipped meanwhile.
+        var stored = LegacyHomedConfig();
+        stored.HomedAppClientId = PrimaryId;
+        _store.Row = stored;
+
+        await _sut.FlipAsync(TenantId, PrimaryId, Actor, "consent-auto-flip");
+
+        VerifyNeverSaved();
+        Assert.Empty(_savedOpsEvents);
+    }
+
+    [Fact]
+    public async Task Flip_skips_a_tenant_being_offboarded()
+    {
+        var tombstone = LegacyHomedConfig();
+        tombstone.Disabled = true;
+        tombstone.DisabledReason = TenantOffboardFunction.OffboardingDisabledReason;
+        _store.Row = tombstone;
+
+        await _sut.FlipAsync(TenantId, PrimaryId, Actor, "manual-ga");
+
+        VerifyNeverSaved();
+        _detectorMock.Verify(x => x.InvalidateTenant(It.IsAny<string>()), Times.Never);
+        Assert.Empty(_savedOpsEvents);
+    }
+
+    [Fact]
+    public async Task Flip_throws_when_every_conditional_write_lost()
+    {
+        _store.Force = TenantConfigUpdateStatus.Conflict;
+
+        await Assert.ThrowsAsync<InvalidOperationException>(
+            () => _sut.FlipAsync(TenantId, PrimaryId, Actor, "manual-ga"));
+
+        Assert.Empty(_savedOpsEvents);
+    }
+
     [Fact]
     public async Task Flip_back_to_legacy_persists_null_homing()
     {
         var config = LegacyHomedConfig();
         config.HomedAppClientId = PrimaryId;
-        _tenantConfigMock.Setup(x => x.GetConfigurationFreshAsync(TenantId)).ReturnsAsync(config);
+        _store.Row = config;
 
         await _sut.FlipAsync(TenantId, null, Actor, "manual-ga");
 
         // The null-homing invariant: legacy is represented as null, never as the legacy GUID.
-        _tenantConfigMock.Verify(x => x.SaveConfigurationAsync(It.Is<TenantConfiguration>(c =>
-            c.HomedAppClientId == null), It.IsAny<string?>(), It.IsAny<string?>()), Times.Once);
+        Assert.Null(Assert.Single(_store.Writes).Row.HomedAppClientId);
         _detectorMock.Verify(x => x.InvalidateTenant(TenantId), Times.Once);
     }
 }

src/Backend/AutopilotMonitor.Functions.Tests/AppInstallIdentityAndEspBlockingTests.cs

@@ -306,7 +306,7 @@ public class AppInstallIdentityAndEspBlockingTests
         };
 
         var root = TestWire.SerializeToElement(await AppsAnalyticsHelper.BuildAnalyticsResponseAsync(
-            summaries, repo.Object, "Company Portal", days: 30));
+            summaries, repo.Object, "Company Portal", AppInstallSources.Ime, days: 30));
 
         Assert.Equal(1, root.GetProperty("collisionExcluded").GetInt32());
         var stats = root.GetProperty("summary");
@@ -325,7 +325,7 @@ public class AppInstallIdentityAndEspBlockingTests
         var summaries = new List<AppInstallSummary> { Summary("Company Portal", collision: true) };
 
         var root = TestWire.SerializeToElement(await AppsAnalyticsHelper.BuildAnalyticsResponseAsync(
-            summaries, repo.Object, "Company Portal", days: 30));
+            summaries, repo.Object, "Company Portal", AppInstallSources.Ime, days: 30));
 
         Assert.Equal(1, root.GetProperty("collisionExcluded").GetInt32());
         Assert.Equal(0, root.GetProperty("summary").GetProperty("totalInstalls").GetInt32());

src/Backend/AutopilotMonitor.Functions.Tests/AppVersionRegressionRadarTests.cs

@@ -195,6 +195,37 @@ public class AppVersionRegressionRadarTests
         Assert.Empty(AppVersionRegressionRadar.Evaluate(rows2));
     }
 
+    [Fact]
+    public void Evaluate_SameNameInTwoChannels_IsTwoApps()
+    {
+        // The RealmJoin package regressed; the Intune app of the same name did not. A shared
+        // group would dilute the RealmJoin medians with the Intune rows and hide the regression.
+        var rows = new List<AppInstallSummary>();
+        rows.AddRange(Batch("1.0", 10, 300, Now.AddDays(-20)).Select(s => { s.Source = AppInstallSources.RealmJoin; return s; }));
+        rows.AddRange(Batch("2.0", 10, 900, Now.AddDays(-5)).Select(s => { s.Source = AppInstallSources.RealmJoin; return s; }));
+        rows.AddRange(Batch("1.0", 30, 300, Now.AddDays(-20)));
+        rows.AddRange(Batch("2.0", 30, 300, Now.AddDays(-5)));
+
+        var finding = Assert.Single(AppVersionRegressionRadar.Evaluate(rows));
+        Assert.Equal(AppInstallSources.RealmJoin, finding.Source);
+        Assert.Equal(App, finding.AppName);
+        Assert.Equal(900, finding.CurrentMedianSeconds);
+    }
+
+    [Fact]
+    public void ShouldReArm_ReadsTheAlertsOwnChannel()
+    {
+        // The IME app still regresses; the RealmJoin episode's version has no measured installs.
+        var rows = new List<AppInstallSummary>();
+        rows.AddRange(Batch("1.0", 10, 300, Now.AddDays(-20)));
+        rows.AddRange(Batch("2.0", 10, 900, Now.AddDays(-5)));
+
+        var rjAlert = Alert();
+        rjAlert.Source = AppInstallSources.RealmJoin;
+        Assert.True(AppVersionRegressionRadar.ShouldReArm(rows, rjAlert));
+        Assert.False(AppVersionRegressionRadar.ShouldReArm(rows, Alert()));
+    }
+
     [Fact]
     public void Evaluate_DeterministicOrdering_LiftDesc_ThenAppNameOrdinal()
     {
@@ -279,13 +310,25 @@ public class AppVersionRegressionRadarTests
     {
         Assert.Equal("appversionregression|contoso app_2|1.0_beta_",
             TableHardwareRejectionNotificationTracker.BuildAppVersionRegressionRowKey(
-                "  Contoso App/2 ", " 1.0#Beta? "));
+                AppInstallSources.Ime, "  Contoso App/2 ", " 1.0#Beta? "));
+    }
+
+    [Fact]
+    public void AppVersionRegressionRowKey_ImeKeepsPreChannelShape_OtherChannelsPrefixTheApp()
+    {
+        // Episodes written before channels existed must keep resolving (re-arm deletes them by key).
+        Assert.Equal("appversionregression|contoso vpn|2.0",
+            TableHardwareRejectionNotificationTracker.BuildAppVersionRegressionRowKey(null, App, "2.0"));
+        Assert.Equal("appversionregression|contoso vpn|2.0",
+            TableHardwareRejectionNotificationTracker.BuildAppVersionRegressionRowKey(string.Empty, App, "2.0"));
+        Assert.Equal("appversionregression|realmjoin:contoso vpn|2.0",
+            TableHardwareRejectionNotificationTracker.BuildAppVersionRegressionRowKey(AppInstallSources.RealmJoin, App, "2.0"));
     }
 
     [Fact]
     public void AppVersionRegressionRowKey_NoPrefixCollisionWithOtherKeyspaces()
     {
-        var key = TableHardwareRejectionNotificationTracker.BuildAppVersionRegressionRowKey(App, "1.0");
+        var key = TableHardwareRejectionNotificationTracker.BuildAppVersionRegressionRowKey(AppInstallSources.RealmJoin, App, "1.0");
         Assert.StartsWith("appversionregression|", key, StringComparison.Ordinal);
         // The prefix range [prefix, "appversionregression}") must not swallow the sibling
         // keyspaces ("ruleregression|", "tpmpss|", "{mfr}|{model}").
@@ -293,6 +336,39 @@ public class AppVersionRegressionRadarTests
         Assert.True(string.CompareOrdinal("tpmpss|", "appversionregression}") > 0);
     }
 
+    [Fact]
+    public void AppVersionRegressionEntity_RealmJoinChannel_RoundTripsWithPrefixedKey()
+    {
+        var alert = new AppVersionRegressionAlert
+        {
+            TenantId = TenantA,
+            AppName = App,
+            Source = AppInstallSources.RealmJoin,
+            CurrentVersion = "2.4.0",
+            PreviousVersion = "2.3.9",
+            FirstNotifiedAt = Now,
+            LastEvaluatedAt = Now,
+        };
+
+        var entity = TableHardwareRejectionNotificationTracker.BuildAppVersionRegressionEntity(TenantA, alert);
+        Assert.Equal("appversionregression|realmjoin:contoso vpn|2.4.0", entity.RowKey);
+        Assert.Equal(AppInstallSources.RealmJoin,
+            TableHardwareRejectionNotificationTracker.MapToAppVersionRegressionAlert(entity).Source);
+    }
+
+    [Fact]
+    public void AppVersionRegressionEntity_RowWithoutSourceColumn_MapsToIme()
+    {
+        var entity = new Azure.Data.Tables.TableEntity(TenantA.ToLowerInvariant(), "appversionregression|contoso vpn|2.4.0")
+        {
+            ["TenantId"] = TenantA,
+            ["AppName"] = App,
+            ["CurrentVersion"] = "2.4.0",
+        };
+        Assert.Equal(AppInstallSources.Ime,
+            TableHardwareRejectionNotificationTracker.MapToAppVersionRegressionAlert(entity).Source);
+    }
+
     [Fact]
     public void AppVersionRegressionEntity_RoundTripsAllFields()
     {
@@ -318,6 +394,7 @@ public class AppVersionRegressionRadarTests
         var mapped = TableHardwareRejectionNotificationTracker.MapToAppVersionRegressionAlert(entity);
         Assert.Equal(TenantA, mapped.TenantId);
         Assert.Equal(App, mapped.AppName);
+        Assert.Equal(AppInstallSources.Ime, mapped.Source);
         Assert.Equal("2.4.0", mapped.CurrentVersion);
         Assert.Equal("2.3.9", mapped.PreviousVersion);
         Assert.Equal(1740, mapped.CurrentMedianSeconds);
@@ -371,7 +448,7 @@ public class AppVersionRegressionRadarTests
         };
 
         var root = TestWire.SerializeToElement(await AppsAnalyticsHelper.BuildAnalyticsResponseAsync(
-            rows, sessionRepo.Object, App, days: 30));
+            rows, sessionRepo.Object, App, AppInstallSources.Ime, days: 30));
 
         var version = root.GetProperty("versionBreakdown")[0];
         Assert.Equal("1.0", version.GetProperty("appVersion").GetString());
@@ -395,12 +472,12 @@ public class AppVersionRegressionRadarTests
 
         // Empty-data early return still surfaces the block (episodes can outlive window data).
         var emptyRoot = TestWire.SerializeToElement(await AppsAnalyticsHelper.BuildAnalyticsResponseAsync(
-            new List<AppInstallSummary>(), sessionRepo.Object, App, days: 30,
+            new List<AppInstallSummary>(), sessionRepo.Object, App, AppInstallSources.Ime, days: 30,
             new List<AppVersionRegressionAlert> { alert }));
         Assert.Equal(1, emptyRoot.GetProperty("versionRegressions").GetArrayLength());
 
         var root = TestWire.SerializeToElement(await AppsAnalyticsHelper.BuildAnalyticsResponseAsync(
-            new List<AppInstallSummary> { Install("1.0", 100, Recent) }, sessionRepo.Object, App, days: 30,
+            new List<AppInstallSummary> { Install("1.0", 100, Recent) }, sessionRepo.Object, App, AppInstallSources.Ime, days: 30,
             new List<AppVersionRegressionAlert> { alert }));
 
         var regression = root.GetProperty("versionRegressions")[0];

src/Backend/AutopilotMonitor.Functions.Tests/AppsAnalyticsFailureRateTests.cs

@@ -131,7 +131,7 @@ public class AppsAnalyticsFailureRateTests
             sessionRepo = mock.Object;
         }
         var result = await AppsAnalyticsHelper.BuildAnalyticsResponseAsync(
-            summaries, sessionRepo, "Contoso App", days: 30);
+            summaries, sessionRepo, "Contoso App", AppInstallSources.Ime, days: 30);
         return TestWire.SerializeToElement(result);
     }
 

src/Backend/AutopilotMonitor.Functions.Tests/AuditLogDeletionExclusionTests.cs

@@ -31,9 +31,9 @@ public class AuditLogDeletionExclusionTests
     public void Global_fanout_filter_includes_deletion_exclusion_only_when_requested()
     {
         var without = TableStorageService.BuildAuditLogFilterWithRowKeyBound(
-            Tenant.ToString(), null, null, lastRowKey: null, excludeDeletions: false);
+            Tenant.ToString(), null, null, rowKeyBound: null, excludeDeletions: false);
         var with = TableStorageService.BuildAuditLogFilterWithRowKeyBound(
-            Tenant.ToString(), null, null, lastRowKey: null, excludeDeletions: true);
+            Tenant.ToString(), null, null, rowKeyBound: null, excludeDeletions: true);
 
         Assert.DoesNotContain("deletion_started", without);
         Assert.Contains("Action ne 'deletion_started'", with);
@@ -47,7 +47,7 @@ public class AuditLogDeletionExclusionTests
         // noise leak back into one of the two views.
         var tenant = TableStorageService.BuildAuditLogFilter(Tenant.ToString(), null, null, excludeDeletions: true);
         var global = TableStorageService.BuildAuditLogFilterWithRowKeyBound(
-            Tenant.ToString(), null, null, lastRowKey: null, excludeDeletions: true);
+            Tenant.ToString(), null, null, rowKeyBound: null, excludeDeletions: true);
 
         var clause = TableStorageService.DeletionExclusionClause();
         Assert.Contains(clause, tenant);

src/Backend/AutopilotMonitor.Functions.Tests/AuditLogFieldFilterTests.cs

@@ -51,7 +51,7 @@ public class AuditLogFieldFilterTests
         var filters = new AuditLogQueryFilters { Action = "device_blocked", EntityType = "Device" };
 
         var f = TableStorageService.BuildAuditLogFilterWithRowKeyBound(
-            Tenant, null, null, lastRowKey: null, excludeDeletions: false, filters);
+            Tenant, null, null, rowKeyBound: null, excludeDeletions: false, filters);
 
         Assert.Contains("Action eq 'device_blocked'", f);
         Assert.Contains("EntityType eq 'Device'", f);

src/Backend/AutopilotMonitor.Functions.Tests/AuthFunctionSideEffectTests.cs

@@ -1,5 +1,9 @@
+using AutopilotMonitor.Functions.Functions.Config;
 using AutopilotMonitor.Functions.Functions.Infrastructure;
 using AutopilotMonitor.Functions.Security;
+using AutopilotMonitor.Functions.Tests.Offboarding;
+using AutopilotMonitor.Shared;
+using AutopilotMonitor.Shared.Models.Offboarding;
 using AutopilotMonitor.Functions.Services;
 using AutopilotMonitor.Shared.DataAccess;
 using AutopilotMonitor.Shared.Models;
@@ -32,6 +36,8 @@ public class AuthFunctionSideEffectTests
     private readonly Mock<IMetricsRepository> _metricsRepoMock;
     private readonly Mock<AutopilotMonitor.Functions.Services.Activation.ITenantAutoApproveEnqueuer> _autoApproveEnqueuerMock;
     private readonly FakeSignalRNotificationService _signalR;
+    private readonly FakeOffboardingAuditRepository _offboardingRepo = new();
+    private readonly StoredTenantConfig _store;
     private readonly AuthFunction _sut;
 
     public AuthFunctionSideEffectTests()
@@ -108,12 +114,12 @@ public class AuthFunctionSideEffectTests
             new Mock<AdminIdentityResolver>(
                 _metricsRepoMock.Object, _tenantConfigMock.Object, Mock.Of<ILogger<AdminIdentityResolver>>()) { CallBase = false }.Object,
             _signalR,
-            adminConfigService.Object);
+            adminConfigService.Object,
+            _offboardingRepo);
 
-        // Default: all fire-and-forget calls succeed
-        _tenantConfigMock
-            .Setup(x => x.SaveConfigurationAsync(It.IsAny<TenantConfiguration>(), It.IsAny<string?>(), It.IsAny<string?>()))
-            .Returns(Task.CompletedTask);
+        // Writes go through UpdateAsync / CreateOrUpdateAsync over one stored row (none by default:
+        // a brand-new tenant). Default: all fire-and-forget calls succeed.
+        _store = _tenantConfigMock.StubWrites(TenantId, row: null);
         _telegramMock
             .Setup(x => x.SendNewTenantSignupAsync(It.IsAny<string>(), It.IsAny<string>()))
             .Returns(Task.CompletedTask);
@@ -202,7 +208,7 @@ public class AuthFunctionSideEffectTests
 
         Assert.True(string.IsNullOrEmpty(config.DomainName));
         Assert.Null(config.OnboardedBy);
-        _tenantConfigMock.Verify(x => x.SaveConfigurationAsync(It.IsAny<TenantConfiguration>(), It.IsAny<string?>(), It.IsAny<string?>()), Times.Never);
+        Assert.Empty(_store.Writes);
         _telegramMock.Verify(x => x.SendNewTenantSignupAsync(It.IsAny<string>(), It.IsAny<string>()), Times.Never);
     }
 
@@ -222,7 +228,11 @@ public class AuthFunctionSideEffectTests
         Assert.Equal(Upn, config.OnboardedBy);
         // The DPA version rides along in the same onboarding write (D-252).
         Assert.Equal(CurrentDpa, config.DpaVersion);
-        _tenantConfigMock.Verify(x => x.SaveConfigurationAsync(config, It.IsAny<string?>(), It.IsAny<string?>()), Times.Once);
+        var (written, source, _) = Assert.Single(_store.Writes);
+        Assert.Equal("contoso.com", written.DomainName);
+        Assert.Equal(Upn, written.OnboardedBy);
+        Assert.Equal(CurrentDpa, written.DpaVersion);
+        Assert.Equal("auth", source);
         _telegramMock.Verify(x => x.SendNewTenantSignupAsync(TenantId, Upn), Times.Once);
         _globalNotificationMock.Verify(x => x.CreateNotificationAsync(
             "preview_signup", "New Tenant Signup",
@@ -237,6 +247,122 @@ public class AuthFunctionSideEffectTests
             It.IsAny<CancellationToken>()), Times.Once);
     }
 
+    // -------------------------------------------------------------------------
+    // Returning tenant (D-288): the offboarding deleted the configuration, the pointer
+    // carries the consumed self-service trial into the first-login write.
+    // -------------------------------------------------------------------------
+
+    private void SeedOffboardingPointer(bool trialConsumed) =>
+        _offboardingRepo.Pointers[TenantId] = (new OffboardingByTenantPointer
+        {
+            PartitionKey = Constants.OffboardingPartitionKeys.ByTenant,
+            RowKey = TenantId,
+            TenantId = TenantId,
+            LatestStatus = "Completed",
+            OffboardCount = 1,
+            TrialConsumed = trialConsumed,
+        }, "\"etag\"");
+
+    private static TenantConfiguration ReturningTenantDefault()
+    {
+        // What auth/me hands over once the offboarding deleted the row: an unpersisted default.
+        var config = TenantConfiguration.CreateDefault(TenantId);
+        Assert.False(config.TrialConsumed);
+        return config;
+    }
+
+    [Fact]
+    public async Task HandleNewTenantDomain_ReturningTenantThatUsedTheTrial_CarriesItIntoTheOnboardingWrite()
+    {
+        SeedOffboardingPointer(trialConsumed: true);
+        var config = ReturningTenantDefault();
+
+        await _sut.HandleNewTenantDomainAsync(config, TenantId, Upn);
+
+        Assert.True(config.TrialConsumed);
+        Assert.Equal(Upn, config.OnboardedBy);
+        var written = Assert.Single(_store.Writes).Row;
+        Assert.True(written.TrialConsumed);
+        Assert.Equal(Upn, written.OnboardedBy);
+        // Re-onboarding itself stays unrestricted: same signup path, auto-approve included.
+        _autoApproveEnqueuerMock.Verify(x => x.EnqueueAsync(
+            It.IsAny<AutopilotMonitor.Functions.Services.Activation.TenantAutoApproveEnvelope>(),
+            It.IsAny<TimeSpan?>(), It.IsAny<CancellationToken>()), Times.Once);
+        // The verify criterion of the finding: the trial gate refuses the returning tenant.
+        var deny = PlanManagementFunction.EvaluateTrialStart(config, DateTime.UtcNow);
+        Assert.Equal(Constants.ApiErrorCodes.TrialAlreadyConsumed, deny?.Code);
+    }
+
+    [Theory]
+    [InlineData(false)] // offboarded without ever using the trial
+    [InlineData(null)]  // never offboarded
+    public async Task HandleNewTenantDomain_NoConsumedTrialOnRecord_LeavesTheTrialAvailable(bool? pointerTrialConsumed)
+    {
+        if (pointerTrialConsumed is { } consumed)
+            SeedOffboardingPointer(consumed);
+        var config = ReturningTenantDefault();
+
+        await _sut.HandleNewTenantDomainAsync(config, TenantId, Upn);
+
+        Assert.False(config.TrialConsumed);
+        Assert.Equal("contoso.com", config.DomainName);
+        Assert.False(Assert.Single(_store.Writes).Row.TrialConsumed);
+    }
+
+    [Fact]
+    public async Task HandleNewTenantDomain_OffboardingLookupFails_WritesNothingAndRetriesOnTheNextLogin()
+    {
+        // An onboarded row without the carry-over would hand the trial back; no write at all instead.
+        _offboardingRepo.ThrowOnPointerRead = new Azure.RequestFailedException(503, "Server busy");
+        var config = ReturningTenantDefault();
+
+        await _sut.HandleNewTenantDomainAsync(config, TenantId, Upn);
+
+        Assert.True(string.IsNullOrEmpty(config.DomainName));
+        Assert.Null(config.OnboardedBy);
+        Assert.Empty(_store.Writes);
+        _telegramMock.Verify(x => x.SendNewTenantSignupAsync(It.IsAny<string>(), It.IsAny<string>()), Times.Never);
+        _autoApproveEnqueuerMock.Verify(x => x.EnqueueAsync(
+            It.IsAny<AutopilotMonitor.Functions.Services.Activation.TenantAutoApproveEnvelope>(),
+            It.IsAny<TimeSpan?>(), It.IsAny<CancellationToken>()), Times.Never);
+    }
+
+    [Fact]
+    public async Task HandleNewTenantDomain_ConcurrentFirstLogin_NeverOverwritesOnboardedBy()
+    {
+        // This login's view still has no domain, but another login seeded the stored row meanwhile:
+        // OnboardedBy is write-once (D-061), and the signup side effects belong to the winner.
+        var stored = DefaultConfig();
+        stored.DomainName = "contoso.com";
+        stored.OnboardedBy = "first@contoso.com";
+        _store.Row = stored;
+        var view = DefaultConfig();
+
+        await _sut.HandleNewTenantDomainAsync(view, TenantId, "second@contoso.com");
+
+        Assert.Empty(_store.Writes);
+        Assert.Equal("first@contoso.com", _store.Row!.OnboardedBy);
+        _telegramMock.Verify(x => x.SendNewTenantSignupAsync(It.IsAny<string>(), It.IsAny<string>()), Times.Never);
+        _autoApproveEnqueuerMock.Verify(x => x.EnqueueAsync(
+            It.IsAny<AutopilotMonitor.Functions.Services.Activation.TenantAutoApproveEnvelope>(),
+            It.IsAny<TimeSpan?>(), It.IsAny<CancellationToken>()), Times.Never);
+    }
+
+    [Fact]
+    public async Task HandleNewTenantDomain_OffboardingTombstone_IsNeverSeeded()
+    {
+        // The offboard click writes a Disabled tombstone even onto a row without a domain.
+        var tombstone = DefaultConfig();
+        tombstone.Disabled = true;
+        tombstone.DisabledReason = AutopilotMonitor.Functions.Functions.Admin.TenantOffboardFunction.OffboardingDisabledReason;
+        _store.Row = tombstone;
+
+        await _sut.HandleNewTenantDomainAsync(DefaultConfig(), TenantId, Upn);
+
+        Assert.Empty(_store.Writes);
+        _telegramMock.Verify(x => x.SendNewTenantSignupAsync(It.IsAny<string>(), It.IsAny<string>()), Times.Never);
+    }
+
     // -------------------------------------------------------------------------
     // HandleAuthClientIdTrackingAsync — the write must mutate a cache-BYPASSING
     // read, never the (possibly stale) cached view. Regression for the prod
@@ -261,16 +387,16 @@ public class AuthFunctionSideEffectTests
         fresh.LastAuthClientId = LegacyAppId;
         fresh.HomedAppClientId = PrimaryAppId; // the concurrent flip that must survive
 
-        _tenantConfigMock.Setup(x => x.GetConfigurationFreshAsync(TenantId)).ReturnsAsync(fresh);
+        _store.Row = fresh;
 
         await _sut.HandleAuthClientIdTrackingAsync(cached, TenantId, $"api://{PrimaryAppId}");
 
-        // The FRESH entity was mutated and saved — the flip survives, the stale view is discarded.
-        _tenantConfigMock.Verify(x => x.SaveConfigurationAsync(fresh, It.IsAny<string?>(), It.IsAny<string?>()), Times.Once);
-        _tenantConfigMock.Verify(x => x.SaveConfigurationAsync(cached, It.IsAny<string?>(), It.IsAny<string?>()), Times.Never);
-        Assert.Equal(PrimaryAppId, fresh.HomedAppClientId);
-        Assert.Equal(PrimaryAppId, fresh.LastAuthClientId);
-        Assert.NotNull(fresh.LastAuthClientIdSince);
+        // The FRESH row was mutated and written — the flip survives, the stale view is discarded.
+        var written = Assert.Single(_store.Writes).Row;
+        Assert.Equal(PrimaryAppId, written.HomedAppClientId);
+        Assert.Equal(PrimaryAppId, written.LastAuthClientId);
+        Assert.NotNull(written.LastAuthClientIdSince);
+        Assert.Null(cached.HomedAppClientId);
     }
 
     [Fact]
@@ -286,11 +412,11 @@ public class AuthFunctionSideEffectTests
         fresh.DomainName = "contoso.com";
         fresh.LastAuthClientId = PrimaryAppId;
 
-        _tenantConfigMock.Setup(x => x.GetConfigurationFreshAsync(TenantId)).ReturnsAsync(fresh);
+        _store.Row = fresh;
 
         await _sut.HandleAuthClientIdTrackingAsync(cached, TenantId, PrimaryAppId);
 
-        _tenantConfigMock.Verify(x => x.SaveConfigurationAsync(It.IsAny<TenantConfiguration>(), It.IsAny<string?>(), It.IsAny<string?>()), Times.Never);
+        Assert.Empty(_store.Writes);
     }
 
     [Fact]
@@ -303,8 +429,9 @@ public class AuthFunctionSideEffectTests
 
         await _sut.HandleAuthClientIdTrackingAsync(cached, TenantId, $"api://{PrimaryAppId}");
 
-        _tenantConfigMock.Verify(x => x.GetConfigurationFreshAsync(It.IsAny<string>()), Times.Never);
-        _tenantConfigMock.Verify(x => x.SaveConfigurationAsync(It.IsAny<TenantConfiguration>(), It.IsAny<string?>(), It.IsAny<string?>()), Times.Never);
+        _tenantConfigMock.Verify(x => x.UpdateAsync(It.IsAny<string>(), It.IsAny<Func<TenantConfiguration, bool>>(),
+            It.IsAny<string>(), It.IsAny<string>()), Times.Never);
+        Assert.Empty(_store.Writes);
     }
 
     private const string CurrentDpa = AutopilotMonitor.Shared.Constants.CurrentDpaVersion;
@@ -321,7 +448,7 @@ public class AuthFunctionSideEffectTests
         await _sut.HandleNewTenantDomainAsync(config, TenantId, Upn);
 
         Assert.Null(config.DpaVersion);
-        _tenantConfigMock.Verify(x => x.SaveConfigurationAsync(It.IsAny<TenantConfiguration>(), It.IsAny<string?>(), It.IsAny<string?>()), Times.Never);
+        Assert.Empty(_store.Writes);
     }
 
     [Fact]
@@ -332,10 +459,12 @@ public class AuthFunctionSideEffectTests
         var config = DefaultConfig();
         config.DomainName = null!;
         config.OnboardedBy = "original.requester@contoso.com";
+        _store.Row = ConfigRepoMockExtensions.Clone(config);
 
         await _sut.HandleNewTenantDomainAsync(config, TenantId, Upn);
 
         Assert.Equal("original.requester@contoso.com", config.OnboardedBy);
+        Assert.Equal("original.requester@contoso.com", Assert.Single(_store.Writes).Row.OnboardedBy);
     }
 
     [Fact]
@@ -346,7 +475,7 @@ public class AuthFunctionSideEffectTests
 
         await _sut.HandleNewTenantDomainAsync(config, TenantId, Upn);
 
-        _tenantConfigMock.Verify(x => x.SaveConfigurationAsync(It.IsAny<TenantConfiguration>(), It.IsAny<string?>(), It.IsAny<string?>()), Times.Never);
+        Assert.Empty(_store.Writes);
         _telegramMock.Verify(x => x.SendNewTenantSignupAsync(It.IsAny<string>(), It.IsAny<string>()), Times.Never);
     }
 
@@ -358,7 +487,7 @@ public class AuthFunctionSideEffectTests
 
         await _sut.HandleNewTenantDomainAsync(config, TenantId, "");
 
-        _tenantConfigMock.Verify(x => x.SaveConfigurationAsync(It.IsAny<TenantConfiguration>(), It.IsAny<string?>(), It.IsAny<string?>()), Times.Never);
+        Assert.Empty(_store.Writes);
     }
 
     [Fact]
@@ -369,7 +498,7 @@ public class AuthFunctionSideEffectTests
 
         await _sut.HandleNewTenantDomainAsync(config, TenantId, "nodomain");
 
-        _tenantConfigMock.Verify(x => x.SaveConfigurationAsync(It.IsAny<TenantConfiguration>(), It.IsAny<string?>(), It.IsAny<string?>()), Times.Never);
+        Assert.Empty(_store.Writes);
     }
 
     [Fact]
@@ -385,8 +514,8 @@ public class AuthFunctionSideEffectTests
         // Should not throw — Telegram is fire-and-forget
         await _sut.HandleNewTenantDomainAsync(config, TenantId, Upn);
 
-        // SaveConfig should still have been called before the fire-and-forget
-        _tenantConfigMock.Verify(x => x.SaveConfigurationAsync(config, It.IsAny<string?>(), It.IsAny<string?>()), Times.Once);
+        // The write still landed before the fire-and-forget
+        Assert.Single(_store.Writes);
     }
 
     // -------------------------------------------------------------------------
@@ -400,14 +529,68 @@ public class AuthFunctionSideEffectTests
         config.Disabled = true;
         config.DisabledReason = "Maintenance";
         config.DisabledUntil = DateTime.UtcNow.AddHours(-1); // expired
+        _store.Row = ConfigRepoMockExtensions.Clone(config);
 
         await _sut.HandleAutoReEnableAsync(config, TenantId);
 
         Assert.False(config.Disabled);
         Assert.Null(config.DisabledReason);
         Assert.Null(config.DisabledUntil);
-        Assert.Equal("System (auto-re-enable)", config.UpdatedBy);
-        _tenantConfigMock.Verify(x => x.SaveConfigurationAsync(config, It.IsAny<string?>(), It.IsAny<string?>()), Times.Once);
+        var written = Assert.Single(_store.Writes).Row;
+        Assert.False(written.Disabled);
+        Assert.Null(written.DisabledUntil);
+        Assert.Equal("System (auto-re-enable)", written.UpdatedBy);
+    }
+
+    [Fact]
+    public async Task HandleAutoReEnable_SuspensionRenewedMeanwhile_StaysSuspended()
+    {
+        // This instance cached the expired suspension; a Global Admin renewed it since.
+        var view = DefaultConfig();
+        view.Disabled = true;
+        view.DisabledReason = "Maintenance";
+        view.DisabledUntil = DateTime.UtcNow.AddHours(-1);
+        var stored = ConfigRepoMockExtensions.Clone(view);
+        stored.DisabledReason = "Abuse review";
+        stored.DisabledUntil = DateTime.UtcNow.AddDays(2);
+        _store.Row = stored;
+
+        await _sut.HandleAutoReEnableAsync(view, TenantId);
+
+        Assert.Empty(_store.Writes);
+        Assert.True(view.IsCurrentlyDisabled());
+        Assert.Equal("Abuse review", view.DisabledReason);
+    }
+
+    [Fact]
+    public async Task HandleAutoReEnable_RowIsTheOffboardingTombstone_StaysSuspended()
+    {
+        var view = DefaultConfig();
+        view.Disabled = true;
+        view.DisabledUntil = DateTime.UtcNow.AddHours(-1);
+        var tombstone = ConfigRepoMockExtensions.Clone(view);
+        tombstone.DisabledReason = AutopilotMonitor.Functions.Functions.Admin.TenantOffboardFunction.OffboardingDisabledReason;
+        tombstone.DisabledUntil = null;
+        _store.Row = tombstone;
+
+        await _sut.HandleAutoReEnableAsync(view, TenantId);
+
+        Assert.Empty(_store.Writes);
+        Assert.True(view.IsCurrentlyDisabled());
+    }
+
+    [Fact]
+    public async Task HandleAutoReEnable_StorageFails_ThisLoginStillPasses()
+    {
+        var view = DefaultConfig();
+        view.Disabled = true;
+        view.DisabledUntil = DateTime.UtcNow.AddHours(-1);
+        _store.ThrowOnRead = new Azure.RequestFailedException(503, "Server busy");
+
+        await _sut.HandleAutoReEnableAsync(view, TenantId);
+
+        Assert.False(view.Disabled);
+        Assert.Empty(_store.Writes);
     }
 
     [Fact]
@@ -420,7 +603,7 @@ public class AuthFunctionSideEffectTests
         await _sut.HandleAutoReEnableAsync(config, TenantId);
 
         Assert.True(config.Disabled);
-        _tenantConfigMock.Verify(x => x.SaveConfigurationAsync(It.IsAny<TenantConfiguration>(), It.IsAny<string?>(), It.IsAny<string?>()), Times.Never);
+        Assert.Empty(_store.Writes);
     }
 
     [Fact]
@@ -431,7 +614,7 @@ public class AuthFunctionSideEffectTests
 
         await _sut.HandleAutoReEnableAsync(config, TenantId);
 
-        _tenantConfigMock.Verify(x => x.SaveConfigurationAsync(It.IsAny<TenantConfiguration>(), It.IsAny<string?>(), It.IsAny<string?>()), Times.Never);
+        Assert.Empty(_store.Writes);
     }
 
     [Fact]
@@ -444,7 +627,7 @@ public class AuthFunctionSideEffectTests
         await _sut.HandleAutoReEnableAsync(config, TenantId);
 
         Assert.True(config.Disabled);
-        _tenantConfigMock.Verify(x => x.SaveConfigurationAsync(It.IsAny<TenantConfiguration>(), It.IsAny<string?>(), It.IsAny<string?>()), Times.Never);
+        Assert.Empty(_store.Writes);
     }
 
     // -------------------------------------------------------------------------

src/Backend/AutopilotMonitor.Functions.Tests/AuthenticationMiddlewareTests.cs

@@ -63,7 +63,7 @@ public class AuthenticationMiddlewareTests
     [InlineData("GET", "/api/progress/sessions/lookup")]   // AuthenticatedUserWithRole
     [InlineData("GET", "/api/sessions")]                   // MemberRead
     [InlineData("GET", "/api/audit/logs")]                 // MemberRead
-    [InlineData("PUT", "/api/config/11111111-1111-1111-1111-111111111111")] // TenantAdminOrGA
+    [InlineData("PATCH", "/api/config/11111111-1111-1111-1111-111111111111/fields")] // TenantAdminOrGA
     [InlineData("GET", "/api/global/sessions")]            // GlobalReadOrDelegatedSubset
     [InlineData("POST", "/api/global/raw/logs")]           // GlobalAdminOnly (KQL proxy)
     [InlineData("GET", "/api/health/detailed")]            // AuthenticatedUser (sub-route is NOT blanket-exempt)

src/Backend/AutopilotMonitor.Functions.Tests/BusinessTimestampFilterClauseTests.cs

@@ -152,7 +152,7 @@ public class BusinessTimestampFilterClauseTests
 
         var plain = TableStorageService.BuildAuditLogFilter(Tenant, from, to);
         var fanOut = TableStorageService.BuildAuditLogFilterWithRowKeyBound(
-            Tenant, from, to, lastRowKey: "!0123_x", excludeDeletions: false);
+            Tenant, from, to, rowKeyBound: new PerPartitionFanOutMerge.RowKeyBound("!0123_x", Inclusive: false), excludeDeletions: false);
 
         foreach (var f in new[] { plain!, fanOut })
         {

src/Backend/AutopilotMonitor.Functions.Tests/CertificateValidatorTests.cs

@@ -1,3 +1,5 @@
+using System.Net;
+using System.Net.Sockets;
 using System.Security.Cryptography;
 using System.Security.Cryptography.X509Certificates;
 using AutopilotMonitor.Functions.Security;
@@ -13,6 +15,7 @@ namespace AutopilotMonitor.Functions.Tests;
 ///     must validate successfully. This guards against rollouts where the embedded root/intermediate
 ///     bundle drifts from the chain Microsoft is actually issuing.
 ///   - Negative path: self-signed leaf certs (with or without Intune-shaped DNs) must NOT validate.
+///   - No network I/O: a leaf from an unknown CA must be rejected without fetching its AIA URL.
 /// </summary>
 public class CertificateValidatorTests
 {
@@ -131,6 +134,31 @@ public class CertificateValidatorTests
         Assert.Contains("expired", result.ErrorMessage, StringComparison.OrdinalIgnoreCase);
     }
 
+    [Fact]
+    public void ValidateCertificate_WithUnknownIssuerAndAiaUrl_NeverFetchesTheIssuer()
+    {
+        // The caller mints the leaf, so the AIA caIssuers URL points wherever they like. A
+        // self-signed leaf never reaches the issuer-download path; it takes a leaf signed by a
+        // CA that is neither embedded nor sent.
+        var listener = new TcpListener(IPAddress.Loopback, 0);
+        listener.Start();
+        try
+        {
+            var port = ((IPEndPoint)listener.LocalEndpoint).Port;
+            var b64 = CreateLeafFromUnknownCaBase64($"http://127.0.0.1:{port}/{Guid.NewGuid():N}.cer");
+
+            var result = CertificateValidator.ValidateCertificate(b64);
+
+            Assert.False(result.IsValid);
+            Assert.Contains("chain validation failed", result.ErrorMessage, StringComparison.OrdinalIgnoreCase);
+            Assert.False(listener.Pending(), "Chain building connected to the certificate's AIA URL");
+        }
+        finally
+        {
+            listener.Stop();
+        }
+    }
+
     private static string ResolveSamplePath()
     {
         var assemblyDir = Path.GetDirectoryName(typeof(CertificateValidatorTests).Assembly.Location)!;
@@ -165,4 +193,24 @@ public class CertificateValidatorTests
 
         return Convert.ToBase64String(cert.Export(X509ContentType.Cert));
     }
+
+    private static string CreateLeafFromUnknownCaBase64(string aiaCaIssuersUrl)
+    {
+        using var caKey = RSA.Create(2048);
+        var caReq = new CertificateRequest("CN=Unknown Test CA", caKey, HashAlgorithmName.SHA256, RSASignaturePadding.Pkcs1);
+        caReq.CertificateExtensions.Add(new X509BasicConstraintsExtension(true, false, 0, true));
+        using var ca = caReq.CreateSelfSigned(DateTimeOffset.UtcNow.AddDays(-1), DateTimeOffset.UtcNow.AddYears(1));
+
+        using var leafKey = RSA.Create(2048);
+        var leafReq = new CertificateRequest("CN=test-device.contoso.example", leafKey, HashAlgorithmName.SHA256, RSASignaturePadding.Pkcs1);
+        leafReq.CertificateExtensions.Add(
+            new X509EnhancedKeyUsageExtension(
+                new OidCollection { new Oid("1.3.6.1.5.5.7.3.2") },
+                critical: false));
+        leafReq.CertificateExtensions.Add(
+            new X509AuthorityInformationAccessExtension(ocspUris: null, caIssuersUris: new[] { aiaCaIssuersUrl }));
+        using var leaf = leafReq.Create(ca, DateTimeOffset.UtcNow.AddHours(-1), DateTimeOffset.UtcNow.AddDays(30), RandomNumberGenerator.GetBytes(16));
+
+        return Convert.ToBase64String(leaf.Export(X509ContentType.Cert));
+    }
 }

src/Backend/AutopilotMonitor.Functions.Tests/CollectLogsQuickConfigOpsEventTests.cs

@@ -40,14 +40,23 @@ public class CollectLogsQuickConfigOpsEventTests
             new TenantConfiguration { TenantId = "t1", DiagnosticsUploadMode = "OnFailure", DiagnosticsUploadDestination = "Hosted" })!;
 
     [Theory]
-    [InlineData("collect-logs", UpdateTenantConfigurationFunction.CollectLogsSource)]
-    [InlineData("COLLECT-LOGS", UpdateTenantConfigurationFunction.CollectLogsSource)]
-    [InlineData(null, "portal-put")]
-    [InlineData("", "portal-put")]
-    [InlineData("something-else", "portal-put")]
-    public void ResolveWriteSource_only_honours_the_allow_listed_intent(string? intent, string expected)
+    [InlineData("?intent=collect-logs", "patch", PatchTenantConfigurationFieldsFunction.CollectLogsSource)]
+    [InlineData("?intent=COLLECT-LOGS", "patch", PatchTenantConfigurationFieldsFunction.CollectLogsSource)]
+    [InlineData("", "patch", "api-patch")]
+    [InlineData("?intent=", "patch", "api-patch")]
+    [InlineData("?intent=something-else", "patch", "api-patch")]
+    [InlineData("?intent=collect-logs", "revert", "api-revert")] // a revert is never the quick-config dialog
+    public void ResolveSource_only_honours_the_allow_listed_intent(string query, string operation, string expected)
     {
-        Assert.Equal(expected, UpdateTenantConfigurationFunction.ResolveWriteSource(intent));
+        var (req, _) = EndpointHarness.Request("t1", queryString: query);
+        Assert.Equal(expected, PatchTenantConfigurationFieldsFunction.ResolveSource(req, operation));
+    }
+
+    [Fact]
+    public void ResolveSource_mcp_header_without_intent_is_mcp()
+    {
+        var (req, _) = EndpointHarness.Request("t1", headers: new Dictionary<string, string> { ["X-Client-Source"] = "mcp" });
+        Assert.Equal("mcp-patch", PatchTenantConfigurationFieldsFunction.ResolveSource(req, "patch"));
     }
 
     [Fact]
@@ -56,7 +65,7 @@ public class CollectLogsQuickConfigOpsEventTests
         var (service, saved) = Rig();
 
         await service.RecordDiagnosticsUploadConfigChangedAsync(
-            "t1", "contoso.com", EnableFlip(), "admin@contoso.com", UpdateTenantConfigurationFunction.CollectLogsSource);
+            "t1", "contoso.com", EnableFlip(), "admin@contoso.com", PatchTenantConfigurationFieldsFunction.CollectLogsSource);
 
         var evt = Assert.Single(saved);
         Assert.Equal("CollectLogsQuickConfigEnabled", evt.EventType);
@@ -85,7 +94,7 @@ public class CollectLogsQuickConfigOpsEventTests
             new TenantConfiguration { TenantId = "t1", DiagnosticsUploadMode = "Always", DiagnosticsUploadDestination = "Hosted" },
             new TenantConfiguration { TenantId = "t1", DiagnosticsUploadMode = "Off", DiagnosticsUploadDestination = "Hosted" })!;
 
-        await service.RecordDiagnosticsUploadConfigChangedAsync("t1", null, disable, "ga@x", UpdateTenantConfigurationFunction.CollectLogsSource);
+        await service.RecordDiagnosticsUploadConfigChangedAsync("t1", null, disable, "ga@x", PatchTenantConfigurationFieldsFunction.CollectLogsSource);
 
         Assert.Equal("DiagnosticsUploadDisabled", Assert.Single(saved).EventType);
     }

src/Backend/AutopilotMonitor.Functions.Tests/ConfigBackupHookTests.cs

@@ -56,6 +56,12 @@ public class ConfigBackupHookTests
                     Upserts.Add(e);
                     return Task.FromResult(Mock.Of<Response>());
                 });
+            Table.Setup(c => c.AddEntityAsync(It.IsAny<TableEntity>(), It.IsAny<CancellationToken>()))
+                .Returns<TableEntity, CancellationToken>((e, _) =>
+                {
+                    Upserts.Add(e);
+                    return Task.FromResult(Mock.Of<Response>());
+                });
             Table.Setup(c => c.UpdateEntityAsync(
                     It.IsAny<TableEntity>(), It.IsAny<ETag>(), It.IsAny<TableUpdateMode>(), It.IsAny<CancellationToken>()))
                 .Returns<TableEntity, ETag, TableUpdateMode, CancellationToken>((e, _, _, _) =>
@@ -109,13 +115,13 @@ public class ConfigBackupHookTests
             .Callback<ConfigBackupEntry, CancellationToken>((e, _) => snapshot = e)
             .Returns(Task.CompletedTask);
 
-        var saved = await harness.Sut.SaveTenantConfigurationAsync(incoming, "portal-put", "test change");
+        var saved = await harness.Sut.TryReplaceTenantConfigurationAsync(incoming, "\"etag\"", "api-patch", "test change");
 
         Assert.True(saved);
         Assert.Single(harness.Upserts); // the config row itself
         Assert.NotNull(snapshot);
         Assert.Equal(TenantId, snapshot!.PartitionKey);
-        Assert.Equal("portal-put", snapshot.Source);
+        Assert.Equal("api-patch", snapshot.Source);
         Assert.Equal("test change", snapshot.Reason);
         Assert.Equal("ga@operator.example", snapshot.ChangedBy); // identity about to write
         // The snapshot holds the STORED row (retention 30), not the incoming one.
@@ -130,11 +136,11 @@ public class ConfigBackupHookTests
     }
 
     [Fact]
-    public async Task RowCreation_NoExistingRow_NoSnapshot()
+    public async Task RowCreation_ConditionalInsert_NoSnapshot()
     {
         var harness = new Harness(stored: null);
 
-        var saved = await harness.Sut.SaveTenantConfigurationAsync(StoredConfig(), "portal-put", null);
+        var saved = await harness.Sut.TryCreateTenantConfigurationAsync(StoredConfig());
 
         Assert.True(saved);
         Assert.Single(harness.Upserts);
@@ -144,6 +150,16 @@ public class ConfigBackupHookTests
             b => b.PruneAsync(It.IsAny<string>(), It.IsAny<int>(), It.IsAny<CancellationToken>()), Times.Never);
     }
 
+    [Fact]
+    public async Task RowCreation_RowAlreadyExists_ReturnsFalse()
+    {
+        var harness = new Harness(stored: null);
+        harness.Table.Setup(c => c.AddEntityAsync(It.IsAny<TableEntity>(), It.IsAny<CancellationToken>()))
+            .ThrowsAsync(new RequestFailedException(409, "Conflict", "EntityAlreadyExists", null));
+
+        Assert.False(await harness.Sut.TryCreateTenantConfigurationAsync(StoredConfig()));
+    }
+
     [Fact]
     public async Task NoiseOnlyChange_AuthClientIdFlip_NoSnapshot()
     {
@@ -157,7 +173,7 @@ public class ConfigBackupHookTests
         incoming.LastUpdated = DateTime.UtcNow;
         incoming.UpdatedBy = "System (auth)";
 
-        var saved = await harness.Sut.SaveTenantConfigurationAsync(incoming, "auth", null);
+        var saved = await harness.Sut.TryReplaceTenantConfigurationAsync(incoming, "\"etag\"", "auth", null);
 
         Assert.True(saved); // the save itself still goes through
         Assert.Single(harness.Upserts);
@@ -171,7 +187,7 @@ public class ConfigBackupHookTests
         var stored = StoredConfig();
         var harness = new Harness(TableConfigRepository.ConvertToTenantTableEntity(stored));
 
-        var saved = await harness.Sut.SaveTenantConfigurationAsync(RoundtripClone(stored), "portal-put", null);
+        var saved = await harness.Sut.TryReplaceTenantConfigurationAsync(RoundtripClone(stored), "\"etag\"", "api-patch", null);
 
         Assert.True(saved);
         harness.Backup.Verify(
@@ -190,14 +206,14 @@ public class ConfigBackupHookTests
         var incoming = RoundtripClone(stored);
         incoming.DataRetentionDays = 90;
 
-        var saved = await harness.Sut.SaveTenantConfigurationAsync(incoming, "portal-put", null);
+        var saved = await harness.Sut.TryReplaceTenantConfigurationAsync(incoming, "\"etag\"", "api-patch", null);
 
         Assert.True(saved);
         Assert.Single(harness.Upserts); // the config write went through regardless
     }
 
     [Fact]
-    public async Task LegacyOneArgOverload_StillSnapshots_WithUnknownSource()
+    public async Task NoSource_StillSnapshots_WithUnknownSource()
     {
         var stored = StoredConfig();
         var harness = new Harness(TableConfigRepository.ConvertToTenantTableEntity(stored));
@@ -210,7 +226,7 @@ public class ConfigBackupHookTests
 
         var incoming = RoundtripClone(stored);
         incoming.DataRetentionDays = 90;
-        await harness.Sut.SaveTenantConfigurationAsync(incoming);
+        await harness.Sut.TryReplaceTenantConfigurationAsync(incoming, "\"etag\"", null, null);
 
         Assert.NotNull(snapshot);
         Assert.Equal("unknown", snapshot!.Source);

src/Backend/AutopilotMonitor.Functions.Tests/ConfigRepoMockExtensions.cs

@@ -0,0 +1,132 @@
+using AutopilotMonitor.Functions.DataAccess.TableStorage;
+using AutopilotMonitor.Functions.Functions.Admin;
+using AutopilotMonitor.Functions.Services;
+using AutopilotMonitor.Shared.DataAccess;
+using AutopilotMonitor.Shared.Models;
+using Moq;
+
+namespace AutopilotMonitor.Functions.Tests;
+
+internal static class ConfigRepoMockExtensions
+{
+    /// <summary>
+    /// Asserts that no tenant configuration row was created or replaced — the only two ways the
+    /// repository can write one (D-290).
+    /// </summary>
+    public static void VerifyNoTenantConfigWrite(this Mock<IConfigRepository> repo)
+    {
+        repo.Verify(r => r.TryCreateTenantConfigurationAsync(It.IsAny<TenantConfiguration>()), Times.Never);
+        repo.Verify(r => r.TryReplaceTenantConfigurationAsync(It.IsAny<TenantConfiguration>(), It.IsAny<string>()), Times.Never);
+        repo.Verify(r => r.TryReplaceTenantConfigurationAsync(
+            It.IsAny<TenantConfiguration>(), It.IsAny<string>(), It.IsAny<string?>(), It.IsAny<string?>()), Times.Never);
+    }
+
+    /// <summary>
+    /// Backs <see cref="TenantConfigurationService.UpdateAsync"/> and
+    /// <see cref="TenantConfigurationService.CreateOrUpdateAsync"/> of a service mock with one stored row
+    /// and the real contract: the mutation sees a fresh copy (never the caller's object), the offboarding
+    /// tombstone is refused, a declined mutation writes nothing, and every write is recorded.
+    /// </summary>
+    public static StoredTenantConfig StubWrites(this Mock<TenantConfigurationService> service, string tenantId, TenantConfiguration? row)
+    {
+        var store = new StoredTenantConfig { Row = row };
+
+        Task<TenantConfigUpdate> Run(Func<TenantConfiguration, bool> mutate, string source, string reason, bool createIfMissing)
+        {
+            if (store.Force is { } forced)
+                return Task.FromResult(new TenantConfigUpdate(forced, null));
+            if (store.ThrowOnRead is { } ex)
+                throw ex;
+
+            TenantConfiguration fresh;
+            if (store.Row == null)
+            {
+                if (!createIfMissing)
+                    return Task.FromResult(new TenantConfigUpdate(TenantConfigUpdateStatus.NotFound, null));
+                fresh = TenantConfiguration.CreateDefault(tenantId);
+            }
+            else
+            {
+                fresh = Clone(store.Row);
+                if (TenantOffboardFunction.IsOffboardingTombstone(fresh))
+                    return Task.FromResult(new TenantConfigUpdate(TenantConfigUpdateStatus.OffboardingInProgress, fresh));
+            }
+
+            if (!mutate(fresh))
+                return Task.FromResult(new TenantConfigUpdate(TenantConfigUpdateStatus.Declined, fresh));
+
+            fresh.LastUpdated = DateTime.UtcNow;
+            store.Row = fresh;
+            store.Writes.Add((fresh, source, reason));
+            return Task.FromResult(new TenantConfigUpdate(TenantConfigUpdateStatus.Updated, fresh));
+        }
+
+        service.Setup(x => x.UpdateAsync(tenantId, It.IsAny<Func<TenantConfiguration, bool>>(), It.IsAny<string>(), It.IsAny<string>()))
+            .Returns((string _, Func<TenantConfiguration, bool> mutate, string source, string reason) => Run(mutate, source, reason, false));
+        service.Setup(x => x.CreateOrUpdateAsync(tenantId, It.IsAny<Func<TenantConfiguration, bool>>(), It.IsAny<string>(), It.IsAny<string>()))
+            .Returns((string _, Func<TenantConfiguration, bool> mutate, string source, string reason) => Run(mutate, source, reason, true));
+        return store;
+    }
+
+    /// <summary>
+    /// Backs the tenant-configuration reads and writes of a repository mock with one stored row and
+    /// the storage contract: every read hands back a fresh copy, the replace succeeds only with the
+    /// current ETag, the insert only while no row exists, and every write is recorded.
+    /// </summary>
+    public static StoredTenantConfig BackWithRow(this Mock<IConfigRepository> repo, TenantConfiguration? row)
+    {
+        var store = new StoredTenantConfig { Row = row };
+        var version = 0;
+        string ETag() => $"etag-{version}";
+
+        TenantConfiguration? Read()
+        {
+            if (store.ThrowOnRead is { } ex) throw ex;
+            return store.Row == null ? null : Clone(store.Row);
+        }
+
+        bool Write(TenantConfiguration config, string source, string reason)
+        {
+            if (store.ThrowOnWrite is { } ex) throw ex;
+            if (store.RejectWrites) return false;
+            store.Row = Clone(config);
+            version++;
+            store.Writes.Add((store.Row, source, reason));
+            return true;
+        }
+
+        repo.Setup(r => r.GetTenantConfigurationAsync(It.IsAny<string>())).ReturnsAsync(() => Read());
+        repo.Setup(r => r.GetTenantConfigurationWithEtagAsync(It.IsAny<string>()))
+            .ReturnsAsync(() => Read() is { } fresh ? (fresh, ETag()) : null);
+        repo.Setup(r => r.TryReplaceTenantConfigurationAsync(
+                It.IsAny<TenantConfiguration>(), It.IsAny<string>(), It.IsAny<string?>(), It.IsAny<string?>()))
+            .ReturnsAsync((TenantConfiguration c, string etag, string? source, string? reason) =>
+                store.Row != null && etag == ETag() && Write(c, source ?? "", reason ?? ""));
+        repo.Setup(r => r.TryCreateTenantConfigurationAsync(It.IsAny<TenantConfiguration>()))
+            .ReturnsAsync((TenantConfiguration c) => store.Row == null && Write(c, "(create)", ""));
+        return store;
+    }
+
+    /// <summary>A storage round trip: what a fresh read of the stored row hands back.</summary>
+    public static TenantConfiguration Clone(TenantConfiguration config)
+        => TableConfigRepository.ConvertFromTenantTableEntity(TableConfigRepository.ConvertToTenantTableEntity(config));
+}
+
+/// <summary>The stored row behind <see cref="ConfigRepoMockExtensions.StubWrites"/> and what was written to it.</summary>
+internal sealed class StoredTenantConfig
+{
+    public TenantConfiguration? Row { get; set; }
+    public List<(TenantConfiguration Row, string Source, string Reason)> Writes { get; } = new();
+
+    /// <summary>When set, every write returns this status without touching the row (e.g. Conflict).</summary>
+    public TenantConfigUpdateStatus? Force { get; set; }
+
+    /// <summary>When set, every read (and, for the service stub, every write) throws it — a storage outage.</summary>
+    public Exception? ThrowOnRead { get; set; }
+
+    /// <summary>Repository double only: when set, every write throws it.</summary>
+    public Exception? ThrowOnWrite { get; set; }
+
+    /// <summary>Repository double only: every conditional write loses (as if a concurrent writer always won).</summary>
+    public bool RejectWrites { get; set; }
+}

src/Backend/AutopilotMonitor.Functions.Tests/ConfigWireParityTests.cs

@@ -713,31 +713,6 @@ public class ConfigWireParityTests
 
     // ---- UpdateTenantConfiguration -------------------------------------------------------
 
-    [Fact]
-    public void UpdateTenantConfigurationResponse_matches_the_ack_plus_config_shape()
-    {
-        var config = new TenantConfiguration
-        {
-            TenantId = "6a6a35a2-30b2-4f2f-9a1b-6d9f1a2b3c4d",
-            DomainName = "contoso.com",
-            UpdatedBy = "admin@contoso.com",
-        };
-
-        AssertParity(
-            new
-            {
-                success = true,
-                message = "Configuration updated successfully",
-                config = config
-            },
-            new UpdateTenantConfigurationResponse
-            {
-                Success = true,
-                Message = "Configuration updated successfully",
-                Config = config
-            });
-    }
-
     private static void AssertParity(object anonymousLiteral, IApiResponse typed)
         => ApiResponseWireParityTests.AssertWireIdentical(anonymousLiteral, typed);
 }

src/Backend/AutopilotMonitor.Functions.Tests/CrossTenantAccessTests.cs

@@ -74,8 +74,7 @@ public class CrossTenantAccessTests
 
     [Theory]
     [InlineData("GET", "/api/config/{0}")]
-    [InlineData("PUT", "/api/config/{0}")]
-    [InlineData("POST", "/api/config/{0}")]
+    [InlineData("PATCH", "/api/config/{0}/fields")]
     [InlineData("GET", "/api/config/{0}/feature-flags")]
     [InlineData("GET", "/api/tenants/{0}/admins")]
     [InlineData("POST", "/api/tenants/{0}/admins")]
@@ -96,8 +95,7 @@ public class CrossTenantAccessTests
 
     [Theory]
     [InlineData("GET", "/api/config/{0}")]
-    [InlineData("PUT", "/api/config/{0}")]
-    [InlineData("POST", "/api/config/{0}")]
+    [InlineData("PATCH", "/api/config/{0}/fields")]
     [InlineData("GET", "/api/config/{0}/feature-flags")]
     [InlineData("GET", "/api/tenants/{0}/admins")]
     [InlineData("POST", "/api/tenants/{0}/admins")]
@@ -118,7 +116,7 @@ public class CrossTenantAccessTests
 
     [Theory]
     [InlineData("GET", "/api/config/{0}")]
-    [InlineData("PUT", "/api/config/{0}")]
+    [InlineData("PATCH", "/api/config/{0}/fields")]
     [InlineData("GET", "/api/tenants/{0}/admins")]
     [InlineData("DELETE", "/api/tenants/{0}/offboard")]
     public void CrossTenant_GlobalAdmin_IsAllowed(string httpMethod, string pathTemplate)

src/Backend/AutopilotMonitor.Functions.Tests/DelegatedAdminServiceTests.cs

@@ -584,17 +584,18 @@ public class DelegatedAdminServiceTests
     }
 
     [Fact]
-    public async Task RenameGroup_DoesNotInvalidateCache()
+    public async Task UpdateGroup_DoesNotInvalidateCache()
     {
         var (svc, repo) = Build();
         ReturnsGroups(repo, Assignment(GroupId1));
         GroupTenants(repo, GroupId1, TenantA);
-        repo.Setup(r => r.RenameTenantGroupAsync(It.IsAny<string>(), It.IsAny<string>())).ReturnsAsync(true);
+        repo.Setup(r => r.UpdateTenantGroupAsync(It.IsAny<string>(), It.IsAny<string?>(), It.IsAny<string?>())).ReturnsAsync(true);
 
         await svc.GetScopeAsync(Id(Upn));
-        await svc.RenameGroupAsync(GroupId1, "Renamed");
-        await svc.GetScopeAsync(Id(Upn)); // name-only change → scope stays cached
+        await svc.UpdateGroupAsync(GroupId1, "Renamed", "Service Desk");
+        await svc.GetScopeAsync(Id(Upn)); // name/label change → scope stays cached
 
+        repo.Verify(r => r.UpdateTenantGroupAsync(GroupId1, "Renamed", "Service Desk"), Times.Once);
         repo.Verify(r => r.GetGroupAssignmentsForUpnAsync(It.IsAny<string>()), Times.Once);
     }
 }

src/Backend/AutopilotMonitor.Functions.Tests/DelegationSelfServiceTests.cs

@@ -563,4 +563,62 @@ public class DelegationSelfServiceTests
         Assert.False(group.Revocable);
         Assert.Equal(2, group.Assignees.Count);
     }
+
+    [Fact]
+    public async Task ListManagers_TwoLabelledOperatorGroups_AreDistinguishable_DirectGrantsStayNeutral()
+    {
+        const string serviceGroupId = "0f1e2d3c4b5a69788796a5b4c3d2e1f0";
+        const string careGroupId = "1a2b3c4d5e6f70819293a4b5c6d7e8f9";
+        var h = Build();
+        h.Repo.Setup(x => x.GetGroupIdsContainingTenantAsync(Customer)).ReturnsAsync(new List<string> { serviceGroupId, careGroupId });
+        h.Repo.Setup(x => x.GetTenantGroupAsync(serviceGroupId)).ReturnsAsync(new TenantGroup
+        {
+            GroupId = serviceGroupId, Name = "Internal managed service tenants", CustomerLabel = "Contoso Managed Service Team",
+            TenantIds = new List<string> { Customer },
+        });
+        h.Repo.Setup(x => x.GetTenantGroupAsync(careGroupId)).ReturnsAsync(new TenantGroup
+        {
+            GroupId = careGroupId, Name = "Internal care unit", CustomerLabel = "  Contoso Customer Care Unit  ",
+            TenantIds = new List<string> { Customer },
+        });
+        h.Repo.Setup(x => x.GetDelegatedAssigneesAsync(Customer)).ReturnsAsync(new List<DelegatedAdminEntry>
+        {
+            new() { Upn = "ops@vendor.example", TenantId = Customer, Role = Constants.DelegatedRoles.DelegatedReader, IsEnabled = true, Status = Constants.DelegatedStatus.Active, GrantedAt = Now.AddDays(-10) },
+        });
+
+        var managers = await h.Svc.ListManagersAsync(Customer);
+
+        Assert.Equal("Contoso Managed Service Team", managers.Single(m => m.GroupId == serviceGroupId).Name);
+        Assert.Equal("Contoso Customer Care Unit", managers.Single(m => m.GroupId == careGroupId).Name);
+        Assert.All(managers.Where(m => m.GroupId != null), m =>
+        {
+            Assert.Equal(DelegationSelfService.SourceOperator, m.Source);
+            Assert.False(m.Revocable);
+            Assert.DoesNotContain("Internal", m.Name, StringComparison.Ordinal);
+        });
+        // A direct grant has no group, hence no label — it keeps the neutral name.
+        Assert.Equal(DelegationSelfService.OperatorLabel, managers.Single(m => m.GroupId == null).Name);
+    }
+
+    // ── Customer-facing group name ──────────────────────────────────────────────
+
+    [Theory]
+    [InlineData(null, "Platform support")]
+    [InlineData("", "Platform support")]
+    [InlineData("   ", "Platform support")]
+    [InlineData(" Contoso Service Desk ", "Contoso Service Desk")]
+    public void CustomerNameOf_OperatorGroup_LabelElseNeutral_NeverTheInternalName(string? label, string expected)
+    {
+        var group = new TenantGroup { GroupId = "0f1e2d3c4b5a69788796a5b4c3d2e1f0", Name = "Internal name", CustomerLabel = label };
+
+        Assert.Equal(expected, DelegationSelfService.CustomerNameOf(group));
+    }
+
+    [Fact]
+    public void CustomerNameOf_OwnedGroup_IsItsName_LabelIgnored()
+    {
+        var group = new TenantGroup { GroupId = GroupId, Name = "partner.example", OwnerTenantId = Home, CustomerLabel = "Ignored" };
+
+        Assert.Equal("partner.example", DelegationSelfService.CustomerNameOf(group));
+    }
 }

src/Backend/AutopilotMonitor.Functions.Tests/EndpointHarness.cs

@@ -0,0 +1,74 @@
+using System.Net;
+using System.Text;
+using System.Text.Json;
+using AutopilotMonitor.Functions.Helpers;
+using AutopilotMonitor.Shared;
+using Azure.Core.Serialization;
+using Microsoft.Azure.Functions.Worker;
+using Microsoft.Azure.Functions.Worker.Http;
+using Microsoft.Extensions.DependencyInjection;
+using Moq;
+
+namespace AutopilotMonitor.Functions.Tests;
+
+/// <summary>
+/// Minimal HTTP endpoint harness: a mocked <see cref="HttpRequestData"/> whose FunctionContext carries the
+/// <see cref="RequestContext"/> the policy middleware would have resolved, serialized with the production
+/// wire settings.
+/// </summary>
+internal static class EndpointHarness
+{
+    public static (HttpRequestData Req, FunctionContext Context) Request(
+        string tenantId, string upn = "admin@contoso.com", string? jsonBody = null, string queryString = "",
+        IDictionary<string, string>? headers = null)
+    {
+        var services = new ServiceCollection();
+        services.AddOptions();
+        services.Configure<WorkerOptions>(o => o.Serializer = new JsonObjectSerializer(ApiJsonOptions.Create()));
+        var provider = services.BuildServiceProvider();
+
+        var items = new Dictionary<object, object>
+        {
+            ["RequestContext"] = new RequestContext
+            {
+                TenantId = tenantId,
+                TargetTenantId = tenantId,
+                UserPrincipalName = upn,
+                IsTenantAdmin = true,
+                UserRole = Constants.TenantRoles.Admin,
+            },
+        };
+        var context = new Mock<FunctionContext>();
+        context.SetupGet(c => c.Items).Returns(items);
+        context.SetupGet(c => c.InstanceServices).Returns(provider);
+
+        var req = new Mock<HttpRequestData>(context.Object);
+        var headerCollection = new HttpHeadersCollection();
+        foreach (var (name, value) in headers ?? new Dictionary<string, string>())
+            headerCollection.Add(name, value);
+        req.SetupGet(r => r.Headers).Returns(headerCollection);
+        // Query parses the Url in the base class (a bare mock would hand back null).
+        req.SetupGet(r => r.Url).Returns(new Uri("https://localhost/api/test" + queryString));
+        req.SetupGet(r => r.Query).CallBase();
+        req.SetupGet(r => r.Body).Returns(new MemoryStream(Encoding.UTF8.GetBytes(jsonBody ?? "")));
+        req.Setup(r => r.CreateResponse()).Returns(() => new FakeResponse(context.Object));
+        return (req.Object, context.Object);
+    }
+
+    /// <summary>The <c>code</c> of an error envelope; null when the body carries none.</summary>
+    public static string? ErrorCode(HttpResponseData response)
+    {
+        response.Body.Position = 0;
+        using var doc = JsonDocument.Parse(response.Body);
+        return doc.RootElement.TryGetProperty("code", out var code) ? code.GetString() : null;
+    }
+
+    private sealed class FakeResponse : HttpResponseData
+    {
+        public FakeResponse(FunctionContext context) : base(context) { }
+        public override HttpStatusCode StatusCode { get; set; } = HttpStatusCode.OK;
+        public override HttpHeadersCollection Headers { get; set; } = new();
+        public override Stream Body { get; set; } = new MemoryStream();
+        public override HttpCookies Cookies { get; } = new Mock<HttpCookies>().Object;
+    }
+}

src/Backend/AutopilotMonitor.Functions.Tests/EndpointPolicyCatalogCompletenessTests.cs

@@ -156,14 +156,17 @@ public class EndpointPolicyCatalogCompletenessTests
     public void SameRoute_DifferentMethods_ResolveToDifferentPolicies()
     {
         var getConfig = EndpointAccessPolicyCatalog.FindPolicy("GET", "/api/config/tenant-1");
-        var putConfig = EndpointAccessPolicyCatalog.FindPolicy("PUT", "/api/config/tenant-1");
+        var patchConfig = EndpointAccessPolicyCatalog.FindPolicy("PATCH", "/api/config/tenant-1/fields");
 
         Assert.NotNull(getConfig);
-        Assert.NotNull(putConfig);
+        Assert.NotNull(patchConfig);
         // GET config read is member-tier (Operators/Viewers get the redacted read-only Settings
-        // view; the handler redacts for every non-admin caller); PUT config write stays Admin/GA only.
+        // view; the handler redacts for every non-admin caller); the field PATCH stays Admin/GA only.
         Assert.Equal(EndpointPolicy.MemberRead, getConfig.Policy);
-        Assert.Equal(EndpointPolicy.TenantAdminOrGA, putConfig.Policy);
+        Assert.Equal(EndpointPolicy.TenantAdminOrGA, patchConfig.Policy);
+        // There is no full-model write of the tenant configuration (D-290).
+        Assert.Null(EndpointAccessPolicyCatalog.FindPolicy("PUT", "/api/config/tenant-1"));
+        Assert.Null(EndpointAccessPolicyCatalog.FindPolicy("POST", "/api/config/tenant-1"));
 
         var getRules = EndpointAccessPolicyCatalog.FindPolicy("GET", "/api/rules/gather");
         var postRules = EndpointAccessPolicyCatalog.FindPolicy("POST", "/api/rules/gather");

src/Backend/AutopilotMonitor.Functions.Tests/GenericWebhookTests.cs

@@ -1,4 +1,5 @@
 using System.Text.Json;
+using AutopilotMonitor.Functions.Helpers;
 using AutopilotMonitor.Functions.Functions.Config;
 using AutopilotMonitor.Functions.Services.Notifications;
 using AutopilotMonitor.Shared.Models;
@@ -140,7 +141,7 @@ public class GenericWebhookTests
         Assert.Empty(config.GetGenericWebhookHeaders());
     }
 
-    // ── UpdateTenantConfigurationFunction.ValidateWebhookCustomHeaders ────
+    // ── TenantConfigValidation.ValidateWebhookCustomHeaders ────
 
     [Theory]
     [InlineData(null)]
@@ -149,7 +150,7 @@ public class GenericWebhookTests
     [InlineData("{\"Authorization\":\"Bearer abc\"}")]
     public void ValidateWebhookCustomHeaders_Accepts_EmptyOrValid(string? json)
     {
-        Assert.Null(UpdateTenantConfigurationFunction.ValidateWebhookCustomHeaders(json));
+        Assert.Null(TenantConfigValidation.ValidateWebhookCustomHeaders(json));
     }
 
     [Theory]
@@ -160,7 +161,7 @@ public class GenericWebhookTests
     [InlineData("{\"X-Key\":\"line1\\nline2\"}", "must not contain line breaks")]
     public void ValidateWebhookCustomHeaders_Rejects_Invalid(string json, string expectedFragment)
     {
-        var error = UpdateTenantConfigurationFunction.ValidateWebhookCustomHeaders(json);
+        var error = TenantConfigValidation.ValidateWebhookCustomHeaders(json);
         Assert.NotNull(error);
         Assert.Contains(expectedFragment, error);
     }

src/Backend/AutopilotMonitor.Functions.Tests/HelloWaitTimeoutConfigTests.cs

@@ -0,0 +1,123 @@
+using AutopilotMonitor.Functions.DataAccess.TableStorage;
+using AutopilotMonitor.Functions.Helpers;
+using AutopilotMonitor.Shared.Models;
+using Xunit;
+
+namespace AutopilotMonitor.Functions.Tests;
+
+/// <summary>
+/// Tenant Hello wait (<see cref="TenantConfiguration.HelloWaitTimeoutSeconds"/>): a changed value
+/// must lie in 30..3600 s. Like the retention cap, only a CHANGED value is checked, so a value
+/// stored before the rule existed never blocks an unrelated save or a backup revert (both
+/// validate the whole model).
+/// </summary>
+public class HelloWaitTimeoutConfigTests
+{
+    private const string TenantId = "11111111-1111-1111-1111-111111111111";
+
+    [Theory]
+    [InlineData(30)]
+    [InlineData(300)]
+    [InlineData(1800)]
+    [InlineData(3600)]
+    public void ValidateModel_accepts_the_supported_range(int seconds)
+    {
+        var candidate = TenantConfiguration.CreateDefault(TenantId);
+        candidate.HelloWaitTimeoutSeconds = seconds;
+
+        Assert.Null(TenantConfigValidation.ValidateModel(candidate, TenantConfiguration.CreateDefault(TenantId), isGlobalAdmin: false));
+    }
+
+    [Theory]
+    [InlineData(29)]
+    [InlineData(3601)]
+    [InlineData(0)]
+    [InlineData(-1)]
+    public void ValidateModel_rejects_a_changed_value_outside_the_range(int seconds)
+    {
+        var candidate = TenantConfiguration.CreateDefault(TenantId);
+        candidate.HelloWaitTimeoutSeconds = seconds;
+
+        var error = TenantConfigValidation.ValidateModel(candidate, TenantConfiguration.CreateDefault(TenantId), isGlobalAdmin: true);
+
+        Assert.Equal("Hello wait timeout must be between 30 and 3600 seconds.", error);
+    }
+
+    [Fact]
+    public void ValidateModel_lets_an_unchanged_out_of_range_value_through()
+    {
+        // Written through the API before the range was validated — an unrelated save (or a
+        // revert to a snapshot holding the same value) must not fail on it.
+        var existing = TenantConfiguration.CreateDefault(TenantId);
+        existing.HelloWaitTimeoutSeconds = 86400;
+        var candidate = TenantConfiguration.CreateDefault(TenantId);
+        candidate.HelloWaitTimeoutSeconds = 86400;
+        candidate.ContactEmail = "it@example.com";
+
+        Assert.Null(TenantConfigValidation.ValidateModel(candidate, existing, isGlobalAdmin: false));
+    }
+
+    [Fact]
+    public void Default_is_the_built_in_five_minutes()
+    {
+        Assert.Equal(300, TenantConfiguration.CreateDefault(TenantId).HelloWaitTimeoutSeconds);
+        Assert.Equal(300, new CollectorConfiguration().HelloWaitTimeoutSeconds);
+        Assert.Equal(300, HelloWaitTimeout.EffectiveSeconds(HelloWaitTimeout.DefaultSeconds));
+    }
+
+    [Fact]
+    public void Stored_old_default_stays_valid_and_keeps_the_built_in_window()
+    {
+        // Every tenant that ever saved the agent settings page has the old 30 s default stored.
+        var existing = TenantConfiguration.CreateDefault(TenantId);
+        existing.HelloWaitTimeoutSeconds = 30;
+        var candidate = TenantConfiguration.CreateDefault(TenantId);
+        candidate.HelloWaitTimeoutSeconds = 30;
+        candidate.ContactEmail = "it@example.com";
+
+        Assert.Null(TenantConfigValidation.ValidateModel(candidate, existing, isGlobalAdmin: false));
+        Assert.Equal(300, HelloWaitTimeout.EffectiveSeconds(30));
+    }
+
+    [Theory]
+    [InlineData(30, 300)]
+    [InlineData(300, 300)]
+    [InlineData(301, 301)]
+    [InlineData(3600, 3600)]
+    [InlineData(86400, 3600)]
+    [InlineData(0, 300)]
+    public void EffectiveSeconds_keeps_the_built_in_window_as_floor_and_caps_at_one_hour(int configured, int expected)
+    {
+        Assert.Equal(expected, HelloWaitTimeout.EffectiveSeconds(configured));
+    }
+
+    [Fact]
+    public void Row_without_the_column_reads_as_the_default()
+    {
+        var entity = TableConfigRepository.ConvertToTenantTableEntity(TenantConfiguration.CreateDefault(TenantId));
+        entity.Remove("HelloWaitTimeoutSeconds");
+
+        Assert.Equal(300, TableConfigRepository.ConvertFromTenantTableEntity(entity).HelloWaitTimeoutSeconds);
+    }
+
+    [Fact]
+    public void Stored_value_is_read_verbatim()
+    {
+        var entity = TableConfigRepository.ConvertToTenantTableEntity(TenantConfiguration.CreateDefault(TenantId));
+        entity["HelloWaitTimeoutSeconds"] = 30;
+
+        Assert.Equal(30, TableConfigRepository.ConvertFromTenantTableEntity(entity).HelloWaitTimeoutSeconds);
+    }
+
+    [Fact]
+    public void Configured_value_roundtrips_store_and_map()
+    {
+        var config = TenantConfiguration.CreateDefault(TenantId);
+        config.HelloWaitTimeoutSeconds = 2400;
+
+        var mapped = TableConfigRepository.ConvertFromTenantTableEntity(
+            TableConfigRepository.ConvertToTenantTableEntity(config));
+
+        Assert.Equal(2400, mapped.HelloWaitTimeoutSeconds);
+    }
+}

src/Backend/AutopilotMonitor.Functions.Tests/McpUserServiceTenantMcpSwitchTests.cs

@@ -153,8 +153,7 @@ public class McpUserServiceTenantMcpSwitchTests
         var result = await _sut.IsAllowedAsync(Upn, HomeTenant, Oid);
 
         Assert.True(result.IsAllowed);
-        _configRepo.Verify(r => r.SaveTenantConfigurationAsync(It.IsAny<TenantConfiguration>()), Times.Never);
-        _configRepo.Verify(r => r.SaveTenantConfigurationAsync(It.IsAny<TenantConfiguration>(), It.IsAny<string?>(), It.IsAny<string?>()), Times.Never);
+        _configRepo.VerifyNoTenantConfigWrite();
     }
 
     [Fact]

src/Backend/AutopilotMonitor.Functions.Tests/NotificationChannelsTests.cs

@@ -218,14 +218,14 @@ public class NotificationChannelsTests
         Assert.Equal("https://second.example/hook", channels[1].Url);
     }
 
-    // ── Save-time validation (UpdateTenantConfigurationFunction) ─────────
+    // ── Save-time validation (TenantConfigValidation) ─────────
 
     [Fact]
     public void ValidateNotificationChannels_NullEmptyOrValidList_PassesThrough()
     {
-        Assert.Null(UpdateTenantConfigurationFunction.ValidateNotificationChannels(null));
-        Assert.Null(UpdateTenantConfigurationFunction.ValidateNotificationChannels(""));
-        Assert.Null(UpdateTenantConfigurationFunction.ValidateNotificationChannels(
+        Assert.Null(TenantConfigValidation.ValidateNotificationChannels(null));
+        Assert.Null(TenantConfigValidation.ValidateNotificationChannels(""));
+        Assert.Null(TenantConfigValidation.ValidateNotificationChannels(
             "[" + ChannelJson("ch-1") + "," + ChannelJson("ch-2", providerType: 10) + "]"));
     }
 
@@ -236,7 +236,7 @@ public class NotificationChannelsTests
     [InlineData("[{\"id\":\"a\",\"providerType\":20,\"url\":\"http://plain.example\"}]", "")] // non-https rejected by SsrfGuard format check
     public void ValidateNotificationChannels_RejectsInvalidEntries(string json, string expectedFragment)
     {
-        var error = UpdateTenantConfigurationFunction.ValidateNotificationChannels(json);
+        var error = TenantConfigValidation.ValidateNotificationChannels(json);
         Assert.NotNull(error);
         if (expectedFragment.Length > 0)
             Assert.Contains(expectedFragment, error);
@@ -245,7 +245,7 @@ public class NotificationChannelsTests
     [Fact]
     public void ValidateNotificationChannels_RejectsDuplicateIds()
     {
-        var error = UpdateTenantConfigurationFunction.ValidateNotificationChannels(
+        var error = TenantConfigValidation.ValidateNotificationChannels(
             "[" + ChannelJson("ch-1") + "," + ChannelJson("ch-1") + "]");
         Assert.NotNull(error);
         Assert.Contains("duplicate", error);
@@ -255,7 +255,7 @@ public class NotificationChannelsTests
     public void ValidateNotificationChannels_RejectsBadPerChannelHeaders()
     {
         var json = "[" + ChannelJson("ch-1", extra: ",\"customHeadersJson\":\"not-json\"") + "]";
-        var error = UpdateTenantConfigurationFunction.ValidateNotificationChannels(json);
+        var error = TenantConfigValidation.ValidateNotificationChannels(json);
         Assert.NotNull(error);
         Assert.Contains("headers", error);
     }
@@ -291,7 +291,7 @@ public class NotificationChannelsTests
     public void ValidateNotificationChannels_RejectsBadPerChannelSigningSecret()
     {
         var json = "[" + ChannelJson("ch-1", extra: ",\"signingSecret\":\"too-short\"") + "]";
-        var error = UpdateTenantConfigurationFunction.ValidateNotificationChannels(json);
+        var error = TenantConfigValidation.ValidateNotificationChannels(json);
         Assert.NotNull(error);
         Assert.Contains("signing secret", error);
     }

src/Backend/AutopilotMonitor.Functions.Tests/Offboarding/FakeOffboardingAuditRepository.cs

@@ -88,9 +88,14 @@ internal sealed class FakeOffboardingAuditRepository : IOffboardingAuditReposito
         return Task.CompletedTask;
     }
 
+    /// <summary>When set, every pointer read throws — a storage outage on the re-onboarding lookup.</summary>
+    public Exception? ThrowOnPointerRead { get; set; }
+
     public Task<(OffboardingByTenantPointer? Pointer, string? ETag)> TryGetByTenantPointerAsync(
         string normalizedTenantId, CancellationToken ct = default)
     {
+        if (ThrowOnPointerRead is { } ex)
+            throw ex;
         if (Pointers.TryGetValue(normalizedTenantId, out var entry))
         {
             return Task.FromResult<(OffboardingByTenantPointer?, string?)>((Clone(entry.Pointer), entry.ETag));
@@ -152,5 +157,6 @@ internal sealed class FakeOffboardingAuditRepository : IOffboardingAuditReposito
         PartitionKey = p.PartitionKey, RowKey = p.RowKey, TenantId = p.TenantId,
         LatestHistoryRowKey = p.LatestHistoryRowKey, LatestStatus = p.LatestStatus,
         LatestUpdatedAt = p.LatestUpdatedAt, OffboardCount = p.OffboardCount,
+        TrialConsumed = p.TrialConsumed,
     };
 }

src/Backend/AutopilotMonitor.Functions.Tests/OpsEventFieldFilterTests.cs

@@ -175,7 +175,8 @@ public class OpsEventFieldFilterTests
         // unfiltered rows — the exact bug this test exists to prevent.
         var filters = new OpsEventQueryFilters { EventType = "AgentEmergencyBreak", MinSeverity = "Error" };
 
-        var fanOut = TableOpsEventRepository.BuildFilterWithRowKeyBound("Agent", null, null, "0009", filters);
+        var fanOut = TableOpsEventRepository.BuildFilterWithRowKeyBound(
+            "Agent", null, null, new PerPartitionFanOutMerge.RowKeyBound("0009", Inclusive: false), filters);
 
         Assert.Contains("EventType eq 'AgentEmergencyBreak'", fanOut);
         Assert.Contains("(Severity eq 'Error' or Severity eq 'Critical')", fanOut);

src/Backend/AutopilotMonitor.Functions.Tests/PerPartitionFanOutMergeTests.cs

@@ -1,325 +1,283 @@
 using AutopilotMonitor.Functions.DataAccess.TableStorage;
-using AutopilotMonitor.Shared.DataAccess;
 using static AutopilotMonitor.Functions.DataAccess.TableStorage.PerPartitionFanOutMerge;
 
 namespace AutopilotMonitor.Functions.Tests;
 
 /// <summary>
-/// Cross-partition merge used by both the global all-category ops-events page
-/// and the global cross-tenant audit-logs page. Without this fan-out, Azure
-/// pages cross-partition queries by (PK asc, RK asc) — so page 1 would be
-/// entirely the alphabetically-first partition, masking newer items from
-/// later-alphabet partitions. The merge picks top-pageSize by timestamp across
-/// all per-partition fetches.
+/// Cross-partition paging behind the global all-category ops-events page and the global
+/// cross-tenant audit-log page. Every page fans out one read per partition and merges by
+/// (RowKey, Partition); the continuation is the watermark of the last returned row. The tests
+/// drive the helper against an in-memory partition store that honours the RowKey bound exactly
+/// like the OData clause does.
 /// </summary>
 public class PerPartitionFanOutMergeTests
 {
-    private static OpsEventEntry Entry(string partition, DateTime ts, string id)
-        => new() { Category = partition, Id = id, Timestamp = ts, EventType = "synthetic" };
-
-    private static (string RowKey, OpsEventEntry Item) Row(DateTime ts, string id)
+    // Reverse-tick RowKey like the storage layer writes it: a smaller string is a newer row.
+    private static string Rk(int minutesAgo, string suffix = "")
     {
-        // Mimics the storage-layer revtick rowkey: smaller string = newer timestamp.
-        var rk = $"{(DateTime.MaxValue.Ticks - ts.Ticks):D19}_{id}";
-        return (rk, Entry(partition: "<set-by-caller>", ts, id));
+        var t0 = new DateTime(2026, 9, 28, 12, 0, 0, DateTimeKind.Utc);
+        return $"{DateTime.MaxValue.Ticks - t0.AddMinutes(-minutesAgo).Ticks:D19}{suffix}";
     }
 
-    private static PartitionFetchResult<OpsEventEntry> Fetched(
-        string partition, params (string RowKey, OpsEventEntry Item)[] rows)
+    private sealed class Store
     {
-        var aligned = rows
-            .Select(r => (r.RowKey, Item: new OpsEventEntry
+        private readonly Dictionary<string, List<string>> _rows;
+        private readonly Func<string, bool> _matches;
+        public List<(string Partition, RowKeyBound? Bound, int Limit)> Calls { get; } = new();
+
+        public Store(Dictionary<string, List<string>> rows, Func<string, bool>? matches = null)
+        {
+            _rows = rows;
+            _matches = matches ?? (_ => true);
+        }
+
+        public IEnumerable<string> Partitions => _rows.Keys;
+
+        public PartitionReader<string> Reader => (partition, bound, limit, _) =>
+        {
+            lock (Calls) Calls.Add((partition, bound, limit));
+            var read = _rows[partition]
+                .Where(_matches)
+                .Where(rk => bound == null
+                    || (bound.Inclusive
+                        ? string.CompareOrdinal(rk, bound.RowKey) >= 0
+                        : string.CompareOrdinal(rk, bound.RowKey) > 0))
+                .OrderBy(rk => rk, StringComparer.Ordinal)
+                .Take(limit)
+                .Select(rk => (rk, $"{partition}/{rk}"))
+                .ToList();
+            return Task.FromResult(read);
+        };
+
+        /// <summary>The full newest-first order the pages must reproduce.</summary>
+        public List<string> Expected() => _rows
+            .SelectMany(kv => kv.Value.Where(_matches).Select(rk => (Partition: kv.Key, RowKey: rk)))
+            .OrderBy(r => r.RowKey, StringComparer.Ordinal)
+            .ThenBy(r => r.Partition, StringComparer.Ordinal)
+            .Select(r => $"{r.Partition}/{r.RowKey}")
+            .ToList();
+    }
+
+    private static async Task<List<List<string>>> WalkAsync(Store store, int pageSize)
+    {
+        var pages = new List<List<string>>();
+        string? continuation = null;
+        do
+        {
+            var page = await FetchPageAsync(store.Partitions, pageSize, continuation, store.Reader);
+            pages.Add(page.Items.ToList());
+            continuation = page.NextRawToken;
+            if (pages.Count > 1000) Assert.Fail("Walk did not terminate");
+        }
+        while (continuation != null);
+        return pages;
+    }
+
+    // ────────── Order ──────────────────────────────────────────────────────
+
+    [Fact]
+    public async Task First_page_is_newest_across_partitions_not_alphabetically_first()
+    {
+        var store = new Store(new()
+        {
+            ["Agent"] = new() { Rk(5) },
+            ["Security"] = new() { Rk(0) },
+            ["Tenant"] = new() { Rk(10) },
+        });
+
+        var page = await FetchPageAsync(store.Partitions, 3, null, store.Reader);
+
+        Assert.Equal(new[] { $"Security/{Rk(0)}", $"Agent/{Rk(5)}", $"Tenant/{Rk(10)}" }, page.Items);
+        Assert.Null(page.NextRawToken);
+    }
+
+    [Fact]
+    public async Task First_page_reads_every_partition_unbounded_with_one_row_of_lookahead()
+    {
+        var store = new Store(new() { ["A"] = new() { Rk(1) }, ["B"] = new(), ["C"] = new() { Rk(2) } });
+
+        await FetchPageAsync(store.Partitions, 15, null, store.Reader);
+
+        Assert.Equal(new[] { "A", "B", "C" }, store.Calls.Select(c => c.Partition).OrderBy(p => p));
+        Assert.All(store.Calls, c => Assert.Null(c.Bound));
+        Assert.All(store.Calls, c => Assert.Equal(16, c.Limit));
+    }
+
+    // ────────── Watermark and bounds ───────────────────────────────────────
+
+    [Fact]
+    public void Bound_is_exclusive_up_to_the_watermark_partition_and_inclusive_after_it()
+    {
+        var w = new Watermark("0000000000000000042", "Maintenance");
+
+        Assert.Null(BoundAfter(null, "Agent"));
+        Assert.Equal(new RowKeyBound(w.RowKey, Inclusive: false), BoundAfter(w, "Agent"));
+        Assert.Equal(new RowKeyBound(w.RowKey, Inclusive: false), BoundAfter(w, "Maintenance"));
+        Assert.Equal(new RowKeyBound(w.RowKey, Inclusive: true), BoundAfter(w, "Security"));
+    }
+
+    [Fact]
+    public void Bound_renders_as_an_escaped_odata_rowkey_clause()
+    {
+        Assert.Equal("RowKey gt '!0042_a'", new RowKeyBound("!0042_a", false).ToODataClause());
+        Assert.Equal("RowKey ge 'it''s'", new RowKeyBound("it's", true).ToODataClause());
+    }
+
+    [Fact]
+    public async Task Equal_rowkeys_in_different_partitions_all_survive_a_page_boundary()
+    {
+        // Ops-event RowKeys are bare reverse ticks, so two categories can share one. With a
+        // RowKey-only watermark the second row would be skipped after the cut.
+        var shared = Rk(3);
+        var store = new Store(new()
+        {
+            ["Agent"] = new() { shared },
+            ["Security"] = new() { shared },
+            ["Tenant"] = new() { shared, Rk(4) },
+        });
+
+        var pages = await WalkAsync(store, pageSize: 1);
+
+        Assert.Equal(store.Expected(), pages.SelectMany(p => p));
+        Assert.Equal(4, pages.Count);
+    }
+
+    // ────────── Exact end of list ──────────────────────────────────────────
+
+    [Fact]
+    public async Task No_continuation_when_the_rows_exactly_fill_the_page()
+    {
+        var store = new Store(new() { ["A"] = new() { Rk(1), Rk(2), Rk(3) }, ["B"] = new() });
+
+        var page = await FetchPageAsync(store.Partitions, 3, null, store.Reader);
+
+        Assert.Equal(3, page.Items.Count);
+        Assert.Null(page.NextRawToken);
+    }
+
+    [Fact]
+    public async Task Continuation_when_one_row_is_left_over()
+    {
+        var store = new Store(new() { ["A"] = new() { Rk(1), Rk(2), Rk(3), Rk(4) } });
+
+        var page = await FetchPageAsync(store.Partitions, 3, null, store.Reader);
+
+        Assert.Equal(3, page.Items.Count);
+        Assert.NotNull(page.NextRawToken);
+    }
+
+    // ────────── Full walks ─────────────────────────────────────────────────
+
+    [Theory]
+    [InlineData(1)]
+    [InlineData(2)]
+    [InlineData(3)]
+    [InlineData(5)]
+    [InlineData(15)]
+    [InlineData(100)]
+    public async Task Walk_returns_every_row_once_in_global_order_and_never_an_empty_page(int pageSize)
+    {
+        // Uneven partitions: one empty, one dominant, interleaved timestamps and a tie.
+        var store = new Store(new()
+        {
+            ["00000000-0000-0000-0000-000000000000"] = new() { Rk(0, "_g0"), Rk(40, "_g1") },
+            ["1a"] = new() { Rk(1, "_a"), Rk(2, "_a"), Rk(3, "_a"), Rk(4, "_a"), Rk(5, "_a"), Rk(6, "_a"), Rk(30, "_a") },
+            ["2b"] = new(),
+            ["3c"] = new() { Rk(2, "_c"), Rk(7, "_c"), Rk(8, "_c") },
+            ["4d"] = new() { Rk(9, "_a") },
+            ["5e"] = new() { Rk(9, "_a"), Rk(50, "_e") },
+        });
+
+        var pages = await WalkAsync(store, pageSize);
+
+        var expected = store.Expected();
+        Assert.Equal(expected, pages.SelectMany(p => p));
+        Assert.All(pages, p => Assert.NotEmpty(p));
+        Assert.All(pages.Take(pages.Count - 1), p => Assert.Equal(pageSize, p.Count));
+    }
+
+    [Fact]
+    public async Task Walk_with_a_server_side_filter_returns_exactly_the_matching_rows()
+    {
+        // The reader applies the filter (like `Action ne 'deletion_started'`), so excluded rows
+        // never take a page slot and never become a watermark.
+        var store = new Store(
+            new()
             {
-                Id = r.Item.Id,
-                Category = partition,
-                EventType = r.Item.EventType,
-                Timestamp = r.Item.Timestamp,
-            }))
-            .ToArray();
-        return new PartitionFetchResult<OpsEventEntry>(partition, aligned);
-    }
+                ["A"] = new() { Rk(1, "_keep"), Rk(2, "_drop"), Rk(3, "_drop"), Rk(4, "_keep") },
+                ["B"] = new() { Rk(2, "_drop"), Rk(5, "_keep"), Rk(6, "_drop") },
+            },
+            matches: rk => rk.EndsWith("_keep", StringComparison.Ordinal));
 
-    // ────────── Merge picks newest-across-partitions, not within-partition ──
+        var pages = await WalkAsync(store, pageSize: 2);
 
-    [Fact]
-    public void First_page_returns_newest_across_partitions_not_alphabetically_first()
-    {
-        // Without the merge, Azure's (PK asc, RK asc) ordering would surface
-        // "Agent" partition items first (alphabetically), even though "Security"
-        // has a newer event. The merge must pick by timestamp, not partition.
-        var t0 = DateTime.UtcNow;
-        var fetched = new[]
-        {
-            Fetched("Agent",    Row(t0.AddMinutes(-5), "agent-1")),
-            Fetched("Security", Row(t0,                "sec-1")),
-            Fetched("Tenant",   Row(t0.AddMinutes(-10), "tenant-1")),
-        };
-
-        var (items, _) = MergeAndAdvance(
-            fetched,
-            new Dictionary<string, PartitionContinuation>(),
-            pageSize: 3,
-            timestampSelector: e => e.Timestamp);
-
-        Assert.Equal(3, items.Count);
-        Assert.Equal("sec-1",    items[0].Id); // newest first
-        Assert.Equal("agent-1",  items[1].Id);
-        Assert.Equal("tenant-1", items[2].Id);
+        Assert.Equal(store.Expected(), pages.SelectMany(p => p));
+        Assert.Equal(3, store.Expected().Count);
     }
 
     [Fact]
-    public void Empty_fetch_drops_partition_from_continuation_map()
+    public async Task Empty_store_yields_one_empty_page_without_continuation()
     {
-        // Wire-format v2 invariant: exhausted partitions are absent from the
-        // continuation, not encoded with `x:true`. Consumers treat "missing"
-        // as exhausted on subsequent pages — that's what shrinks the audit
-        // fan-out token from 30+ KB to a few hundred bytes.
-        var t0 = DateTime.UtcNow;
-        var fetched = new[]
-        {
-            Fetched("Agent"),                                 // empty fetch
-            Fetched("Security", Row(t0, "sec-1")),
-        };
+        var store = new Store(new() { ["A"] = new(), ["B"] = new() });
 
-        var (_, conts) = MergeAndAdvance(
-            fetched,
-            new Dictionary<string, PartitionContinuation>(),
-            pageSize: 5,
-            timestampSelector: e => e.Timestamp);
+        var page = await FetchPageAsync(store.Partitions, 15, null, store.Reader);
 
-        Assert.False(conts.ContainsKey("Agent"));
-        Assert.True(conts.ContainsKey("Security"));
+        Assert.Empty(page.Items);
+        Assert.Null(page.NextRawToken);
     }
 
-    // ────────── Continuation advance: avoid losing leftovers when a partition loses the cut ──
+    // ────────── Token ──────────────────────────────────────────────────────
 
     [Fact]
-    public void Continuation_keeps_position_for_partitions_whose_items_lost_the_merge_cut()
+    public async Task Continuation_size_does_not_grow_with_the_number_of_partitions()
     {
-        // Page 1: pageSize=3. Agent fetched 5 items, Security fetched 5 items, but
-        // Security's are all newer. Merge takes 3 from Security; Agent's 5 items
-        // all lost the cut. Agent's continuation MUST stay at its previous
-        // position so the items get re-fetched and compete on page 2.
-        var t0 = DateTime.UtcNow;
-        var fetched = new[]
-        {
-            Fetched("Agent",
-                Row(t0.AddMinutes(-100), "a1"),
-                Row(t0.AddMinutes(-101), "a2"),
-                Row(t0.AddMinutes(-102), "a3"),
-                Row(t0.AddMinutes(-103), "a4"),
-                Row(t0.AddMinutes(-104), "a5")),
-            Fetched("Security",
-                Row(t0,                  "s1"),
-                Row(t0.AddMinutes(-1),   "s2"),
-                Row(t0.AddMinutes(-2),   "s3"),
-                Row(t0.AddMinutes(-3),   "s4"),
-                Row(t0.AddMinutes(-4),   "s5")),
-        };
+        // The per-partition map this replaced grew with every partition holding rows until the
+        // nextLink failed with HTTP 414. Every partition here has rows that lose the merge cut.
+        static Store Tenants(int count) => new(Enumerable.Range(0, count).ToDictionary(
+            i => Guid.Parse($"00000000-0000-0000-0000-{i:D12}").ToString(),
+            i => new List<string> { Rk(i, "_" + Guid.Empty.ToString("N")), Rk(i + 1000, "_" + Guid.Empty.ToString("N")) }));
 
-        var (items, conts) = MergeAndAdvance(
-            fetched,
-            new Dictionary<string, PartitionContinuation>(),
-            pageSize: 3,
-            timestampSelector: e => e.Timestamp);
+        var few = await FetchPageAsync(Tenants(3).Partitions, 1, null, Tenants(3).Reader);
+        var many = await FetchPageAsync(Tenants(500).Partitions, 1, null, Tenants(500).Reader);
 
-        // Security wins all three slots.
-        Assert.Equal(new[] { "s1", "s2", "s3" }, items.Select(i => i.Id).ToArray());
-
-        // Agent had no items returned and a non-empty fetch → keep prior continuation
-        // (null = restart). The partition stays in the map (active with no rk),
-        // so the next page will re-fetch its items and they'll be re-considered
-        // at a lower floor.
-        Assert.True(conts.ContainsKey("Agent"));
-        Assert.Null(conts["Agent"].LastRowKey);
-
-        Assert.True(conts.ContainsKey("Security"));
-        Assert.NotNull(conts["Security"].LastRowKey);
+        Assert.NotNull(many.NextRawToken);
+        Assert.Equal(few.NextRawToken!.Length, many.NextRawToken!.Length);
+        Assert.True(many.NextRawToken.Length < 200, $"raw continuation is {many.NextRawToken.Length} characters");
     }
 
     [Fact]
-    public void Continuation_advances_to_oldest_returned_rowkey_per_partition()
+    public void Watermark_round_trips()
     {
-        // When merge takes some-but-not-all from a partition, advance only past the
-        // items we actually returned. The leftover (older) items in that fetch
-        // will be re-fetched and re-considered on the next page.
-        var t0 = DateTime.UtcNow;
-        var (rkNew, _) = Row(t0,                "x-new");
-        var (rkMid, _) = Row(t0.AddMinutes(-5), "x-mid");
-        var (rkOld, _) = Row(t0.AddMinutes(-10), "x-old");
-        var fetched = new[]
-        {
-            Fetched("Agent",
-                (rkNew, Entry("Agent", t0,                "x-new")),
-                (rkMid, Entry("Agent", t0.AddMinutes(-5), "x-mid")),
-                (rkOld, Entry("Agent", t0.AddMinutes(-10), "x-old"))),
-            // A second partition whose only event is between Agent's mid and old
-            // items, ensuring the merge ends after Agent's mid item.
-            Fetched("Security",
-                Row(t0.AddMinutes(-7), "s-only")),
-        };
+        var w = new Watermark("!2516116980007133310_a4bb51498cad468f89431b15f1e701f7", "tenant-a");
 
-        var (items, conts) = MergeAndAdvance(
-            fetched,
-            new Dictionary<string, PartitionContinuation>(),
-            pageSize: 3,
-            timestampSelector: e => e.Timestamp);
-
-        // Top 3 by timestamp desc: x-new, x-mid, s-only. x-old is dropped.
-        Assert.Equal(new[] { "x-new", "x-mid", "s-only" }, items.Select(i => i.Id).ToArray());
-
-        // Agent's continuation advances to rkMid (the oldest Agent rowkey we returned).
-        // x-old will be re-fetched and re-considered next page (filter: RowKey gt rkMid).
-        Assert.Equal(rkMid, conts["Agent"].LastRowKey);
-    }
-
-    // ────────── Codec round-trip ────────────────────────────────────────────
-
-    [Fact]
-    public void Encode_then_decode_round_trips_active_partitions()
-    {
-        var original = new Dictionary<string, PartitionContinuation>
-        {
-            ["Agent"]   = new PartitionContinuation("0123456789012345678_aaa"),
-            ["Consent"] = new PartitionContinuation("0987654321098765432_bbb"),
-            ["Tenant"]  = new PartitionContinuation(null), // active, restart from top
-        };
-
-        var encoded = EncodeMultiContinuation(original);
-        var decoded = DecodeMultiContinuation(encoded);
-
-        Assert.Equal(3, decoded.Count);
-        Assert.Equal("0123456789012345678_aaa", decoded["Agent"].LastRowKey);
-        Assert.Equal("0987654321098765432_bbb", decoded["Consent"].LastRowKey);
-        Assert.Null(decoded["Tenant"].LastRowKey);
+        Assert.Equal(w, DecodeWatermark(EncodeWatermark(w)));
     }
 
     [Fact]
-    public void Decode_silently_drops_legacy_v1_exhausted_entries()
+    public void Decode_rejects_anything_that_is_not_a_v3_watermark()
     {
-        // Wire format v1 (pre-fix) explicitly encoded `x:true` for exhausted
-        // partitions; v2 just omits them. Tokens that were already in flight
-        // when the deploy went out must continue to page through cleanly —
-        // those legacy `x:true` entries are dropped silently so the
-        // consumer's "missing = exhausted" rule applies.
-        var legacyJson = """{"v":1,"c":{"Agent":{"rk":"0123456789012345678_aaa","x":false},"Security":{"rk":null,"x":true},"Consent":{"rk":"0987654321098765432_bbb","x":false}}}""";
-        var legacyToken = Convert.ToBase64String(System.Text.Encoding.UTF8.GetBytes(legacyJson));
+        static string B64(string json) => Convert.ToBase64String(System.Text.Encoding.UTF8.GetBytes(json));
 
-        var decoded = DecodeMultiContinuation(legacyToken);
-
-        Assert.Equal(2, decoded.Count);
-        Assert.True(decoded.ContainsKey("Agent"));
-        Assert.True(decoded.ContainsKey("Consent"));
-        Assert.False(decoded.ContainsKey("Security")); // dropped: was x:true in v1 wire
+        Assert.Null(DecodeWatermark(null));
+        Assert.Null(DecodeWatermark(""));
+        Assert.Null(DecodeWatermark("not-base64-!@#"));
+        Assert.Null(DecodeWatermark(B64("this is not json")));
+        Assert.Null(DecodeWatermark(B64("""{"v":2,"c":{"Agent":{"rk":"0123"}}}""")));
+        Assert.Null(DecodeWatermark(B64("""{"v":3,"rk":"","p":"Agent"}""")));
+        Assert.Null(DecodeWatermark(B64("""{"v":3,"rk":"0123"}""")));
+        Assert.Null(DecodeWatermark(B64("""{"v":"3","rk":"0123","p":"Agent"}""")));
     }
 
     [Fact]
-    public void Encode_omits_partition_state_field_to_keep_token_compact()
+    public async Task A_token_from_an_older_deploy_ends_the_list_without_reading_storage()
     {
-        // Sanity-check the wire compactness gain: 200 partitions of v1 with
-        // `x:false` per entry produced ~30 KB after base64; v2 should be
-        // dramatically smaller because we drop the redundant flag and only
-        // encode active entries.
-        var many = Enumerable.Range(0, 200)
-            .ToDictionary(i => $"tenant-{i:D3}", i => new PartitionContinuation($"!1234567890123456789_{i:D8}"));
+        var store = new Store(new() { ["Agent"] = new() { Rk(1) } });
+        var v2 = Convert.ToBase64String(System.Text.Encoding.UTF8.GetBytes("""{"v":2,"c":{"Agent":{"rk":null}}}"""));
 
-        var encoded = EncodeMultiContinuation(many);
+        var page = await FetchPageAsync(store.Partitions, 15, v2, store.Reader);
 
-        // 200 entries × (~30 chars name + ~30 chars rk + JSON overhead) → JSON
-        // ~12 KB → base64 ~16 KB. The legacy `x:false` flag adds another
-        // ~10 bytes per entry (~2 KB total) which we now save. The hard
-        // assertion below is on absolute size: we must stay well below the
-        // ~30 KB that broke audit Next clicks.
-        Assert.True(encoded.Length < 20_000,
-            $"Encoded continuation should stay under 20 KB even at 200 partitions; got {encoded.Length} bytes");
-    }
-
-    [Fact]
-    public void Decode_returns_empty_for_malformed_or_empty_continuation()
-    {
-        Assert.Empty(DecodeMultiContinuation(null));
-        Assert.Empty(DecodeMultiContinuation(""));
-        Assert.Empty(DecodeMultiContinuation("not-base64-!@#"));
-        Assert.Empty(DecodeMultiContinuation("dGhpcyBpcyBub3QganNvbg==")); // valid base64, not JSON
-    }
-
-    // ────────── End-to-end: drain across pages preserves all items, in order ──
-
-    [Fact]
-    public void Multi_page_drain_returns_all_items_globally_newest_first_no_duplicates()
-    {
-        var t0 = new DateTime(2026, 5, 5, 12, 0, 0, DateTimeKind.Utc);
-
-        var allItems = new List<(string Part, OpsEventEntry Item, string Rk)>();
-        var partitions = new[] { "Agent", "Security", "Tenant" };
-        var minute = 0;
-        foreach (var part in partitions)
-        {
-            for (int i = 0; i < 4; i++)
-            {
-                minute++;
-                var ts = t0.AddMinutes(-minute);
-                var (rk, _) = Row(ts, $"{part}-{i}");
-                allItems.Add((part, new OpsEventEntry
-                {
-                    Category = part,
-                    Id = $"{part}-{i}",
-                    Timestamp = ts,
-                    EventType = "synthetic",
-                }, rk));
-            }
-        }
-
-        IReadOnlyList<PartitionFetchResult<OpsEventEntry>> SimulateFetch(
-            Dictionary<string, PartitionContinuation> continuations, int pageSize, bool isFirstPage)
-        {
-            var fetched = new List<PartitionFetchResult<OpsEventEntry>>();
-            foreach (var part in partitions)
-            {
-                // Mirrors the consumer rule: on first page query everything; on
-                // subsequent pages only partitions still in the continuation.
-                if (!isFirstPage && !continuations.ContainsKey(part)) continue;
-                continuations.TryGetValue(part, out var prior);
-                var lastRk = prior?.LastRowKey;
-
-                var rows = allItems
-                    .Where(t => t.Part == part)
-                    .Where(t => lastRk == null || string.Compare(t.Rk, lastRk, StringComparison.Ordinal) > 0)
-                    .OrderBy(t => t.Rk, StringComparer.Ordinal)
-                    .Take(pageSize)
-                    .Select(t => (t.Rk, t.Item))
-                    .ToList();
-                fetched.Add(new PartitionFetchResult<OpsEventEntry>(part, rows));
-            }
-            return fetched;
-        }
-
-        const int pageSize = 5;
-        var seen = new List<OpsEventEntry>();
-        var conts = new Dictionary<string, PartitionContinuation>();
-        var loopGuard = 0;
-        bool isFirstPage = true;
-
-        while (true)
-        {
-            var fetched = SimulateFetch(conts, pageSize, isFirstPage);
-            var (page, nextConts) = MergeAndAdvance(fetched, conts, pageSize, e => e.Timestamp);
-            seen.AddRange(page);
-            conts = nextConts;
-            isFirstPage = false;
-
-            // Map only carries active partitions now — empty == drain complete.
-            if (conts.Count == 0) break;
-            if (page.Count == 0) break;
-            if (++loopGuard > 50) Assert.Fail("Drain loop did not terminate");
-        }
-
-        Assert.Equal(12, seen.Count);
-        Assert.Equal(12, seen.Select(e => e.Id).Distinct().Count());
-        for (int i = 1; i < seen.Count; i++)
-        {
-            Assert.True(seen[i - 1].Timestamp >= seen[i].Timestamp,
-                $"Out-of-order at index {i}: {seen[i - 1].Id}({seen[i - 1].Timestamp}) before {seen[i].Id}({seen[i].Timestamp})");
-        }
+        Assert.Empty(page.Items);
+        Assert.Null(page.NextRawToken);
+        Assert.Empty(store.Calls);
     }
 }

src/Backend/AutopilotMonitor.Functions.Tests/PolicyEnforcementMiddlewareTests.cs

@@ -346,7 +346,7 @@ public class PolicyEnforcementMiddlewareTests
         h.AsTenantAdmin(TenantA, upn);
 
         // PUT config/{tenantId} is TenantAdminOrGA (write). Route tenant == JWT tenant ⇒ own-tenant write.
-        var result = await h.Middleware.DecideAsync("PUT", $"/api/config/{TenantA}", null, AuthedPrincipal(TenantA, upn));
+        var result = await h.Middleware.DecideAsync("PATCH", $"/api/config/{TenantA}/fields", null, AuthedPrincipal(TenantA, upn));
 
         Assert.True(result.Allowed); // additive: the tenant-admin hat still grants write in the own tenant
     }
@@ -361,7 +361,7 @@ public class PolicyEnforcementMiddlewareTests
         h.AsGlobalRole(Constants.GlobalRoles.GlobalReader);
         // No tenant membership ⇒ no write path.
 
-        var result = await h.Middleware.DecideAsync("PUT", $"/api/config/{TenantA}", null, AuthedPrincipal(TenantA, upn));
+        var result = await h.Middleware.DecideAsync("PATCH", $"/api/config/{TenantA}/fields", null, AuthedPrincipal(TenantA, upn));
 
         Assert.False(result.Allowed);
         Assert.Equal(403, result.StatusCode);
@@ -700,7 +700,7 @@ public class PolicyEnforcementMiddlewareTests
         h.AsDelegated(TenantB, Constants.DelegatedRoles.DelegatedAdmin); // even a DelegatedAdmin row…
 
         // PUT config/{B} is TenantAdminOrGA (write). Phase 1: delegation never satisfies a write tier.
-        var result = await h.Middleware.DecideAsync("PUT", $"/api/config/{TenantB}", null, AuthedPrincipal(TenantA, upn));
+        var result = await h.Middleware.DecideAsync("PATCH", $"/api/config/{TenantB}/fields", null, AuthedPrincipal(TenantA, upn));
 
         Assert.False(result.Allowed);
         Assert.Equal(403, result.StatusCode);
@@ -986,9 +986,7 @@ public class PolicyEnforcementMiddlewareTests
 
         Assert.True(result.Allowed); // the delegated read itself is allowed…
         // …but no config row was persisted for the MSP's home tenant A (nor any tenant).
-        h.ConfigRepo.Verify(r => r.SaveTenantConfigurationAsync(It.IsAny<TenantConfiguration>()), Times.Never);
-        h.ConfigRepo.Verify(r => r.SaveTenantConfigurationAsync(
-            It.IsAny<TenantConfiguration>(), It.IsAny<string?>(), It.IsAny<string?>()), Times.Never);
+        h.ConfigRepo.VerifyNoTenantConfigWrite();
     }
 
     // ── Phase 2a: delegated single-tenant access to cross-tenant /api/global/* read endpoints ──
@@ -1547,7 +1545,7 @@ public class PolicyEnforcementMiddlewareTests
         h.AsTenantAdmin(TenantA, upn);
         AsSuspended(h, TenantA); // indefinite suspension (no DisabledUntil)
 
-        var result = await h.Middleware.DecideAsync("PUT", $"/api/config/{TenantA}", null, AuthedPrincipal(TenantA, upn));
+        var result = await h.Middleware.DecideAsync("PATCH", $"/api/config/{TenantA}/fields", null, AuthedPrincipal(TenantA, upn));
 
         Assert.False(result.Allowed);
         Assert.Equal("TenantSuspended", result.ErrorCode);
@@ -1576,7 +1574,7 @@ public class PolicyEnforcementMiddlewareTests
         AsSuspended(h, TenantB);
 
         var own = await h.Middleware.DecideAsync("GET", "/api/sessions", null, AuthedPrincipal(TenantA, upn));
-        var cross = await h.Middleware.DecideAsync("PUT", $"/api/config/{TenantB}", null, AuthedPrincipal(TenantA, upn));
+        var cross = await h.Middleware.DecideAsync("PATCH", $"/api/config/{TenantB}/fields", null, AuthedPrincipal(TenantA, upn));
 
         Assert.True(own.Allowed);
         Assert.True(cross.Allowed);

src/Backend/AutopilotMonitor.Functions.Tests/ProConferralServiceTests.cs

@@ -43,8 +43,10 @@ public sealed class ProConferralServiceTests
         var configRepo = new Mock<IConfigRepository>();
         configRepo.Setup(c => c.GetTenantConfigurationAsync(It.IsAny<string>()))
             .ReturnsAsync((string id) => rows.TryGetValue(id, out var cfg) ? cfg : null);
-        configRepo.Setup(c => c.SaveTenantConfigurationAsync(It.IsAny<TenantConfiguration>(), It.IsAny<string?>(), It.IsAny<string?>()))
-            .ReturnsAsync((TenantConfiguration cfg, string? source, string? reason) => { saves.Add((cfg, source, reason)); return true; });
+        configRepo.Setup(c => c.GetTenantConfigurationWithEtagAsync(It.IsAny<string>()))
+            .ReturnsAsync((string id) => rows.TryGetValue(id, out var cfg) ? (cfg, "etag") : null);
+        configRepo.Setup(c => c.TryReplaceTenantConfigurationAsync(It.IsAny<TenantConfiguration>(), "etag", It.IsAny<string?>(), It.IsAny<string?>()))
+            .ReturnsAsync((TenantConfiguration cfg, string _, string? source, string? reason) => { saves.Add((cfg, source, reason)); return true; });
 
         var adminRepo = new Mock<IAdminRepository>();
         adminRepo.Setup(a => a.GetGroupTenantsAsync(Constants.TenantGroupIds.ForHomeTenant(Home))).ReturnsAsync(members.ToList());

src/Backend/AutopilotMonitor.Functions.Tests/RealmJoinPackageInstallRowsTests.cs

@@ -0,0 +1,314 @@
+using System.Text.Json;
+using Azure.Data.Tables;
+using AutopilotMonitor.Functions.Functions.Apps;
+using AutopilotMonitor.Functions.Helpers;
+using AutopilotMonitor.Functions.Services;
+using AutopilotMonitor.Shared;
+using AutopilotMonitor.Shared.DataAccess;
+using AutopilotMonitor.Shared.Models;
+using Microsoft.Extensions.Logging.Abstractions;
+using Moq;
+using Xunit;
+
+namespace AutopilotMonitor.Functions.Tests;
+
+/// <summary>
+/// RealmJoin packages as <c>AppInstallSummaries</c> rows (<see cref="AppInstallSources.RealmJoin"/>):
+/// the ingest fold of <c>realmjoin_package_started/completed</c>, the scope + package-id row
+/// identity, the store-side channel columns, and the channel dimension of the per-app aggregates
+/// (an Intune app and a RealmJoin package with the same display name never merge).
+/// </summary>
+public class RealmJoinPackageInstallRowsTests
+{
+    private static readonly DateTime T0 = new(2026, 9, 17, 8, 29, 26, DateTimeKind.Utc);
+    private const string SessionId = "11111111-2222-3333-4444-555555555555";
+
+    private static Dictionary<string, AppInstallAggregationState> Aggregate(params EnrollmentEvent[] events)
+    {
+        // Same comparer as the production classification (EventIngestProcessor.Classification).
+        var summaries = new Dictionary<string, AppInstallAggregationState>(StringComparer.OrdinalIgnoreCase);
+        foreach (var evt in events)
+            EventIngestProcessor.AggregateAppInstallEvent(evt, "tenant", SessionId, summaries);
+        return summaries;
+    }
+
+    private static AppInstallSummary Row(Dictionary<string, AppInstallAggregationState> summaries, string packageId, string scope = "machine")
+        => summaries[EventIngestProcessor.RealmJoinAggregationKey(scope, packageId)].Summary;
+
+    private static EnrollmentEvent Started(DateTime at, string packageId = "generic-github-desktop",
+        string displayName = "GitHub Desktop", string scope = "machine", string? version = "3.6.5.0")
+    {
+        var data = new Dictionary<string, object>
+        {
+            ["packageId"] = packageId,
+            ["displayName"] = displayName,
+            ["scope"] = scope,
+        };
+        if (version != null) data["version"] = version;
+        return new EnrollmentEvent { EventType = Constants.EventTypes.RealmJoinPackageStarted, Timestamp = at, Data = data };
+    }
+
+    private static EnrollmentEvent Completed(DateTime at, bool success = true, int exitCode = 0,
+        string packageId = "generic-github-desktop", string displayName = "GitHub Desktop", string scope = "machine")
+    {
+        var evt = Started(at, packageId, displayName, scope);
+        evt.EventType = Constants.EventTypes.RealmJoinPackageCompleted;
+        // Wire form of the agent adapter: strings.
+        evt.Data!["success"] = success ? "true" : "false";
+        evt.Data["lastExitCode"] = exitCode.ToString();
+        return evt;
+    }
+
+    // ── ingest fold: live watcher events ────────────────────────────────────
+
+    [Fact]
+    public void StartedThenCompleted_OneRealmJoinRow_WithMeasuredDuration()
+    {
+        var s = Row(Aggregate(Started(T0), Completed(T0.AddSeconds(22))), "generic-github-desktop");
+
+        Assert.Equal(AppInstallSources.RealmJoin, s.Source);
+        Assert.Equal("generic-github-desktop", s.AppId);
+        Assert.Equal("machine", s.InstallScope);
+        Assert.Equal("GitHub Desktop", s.AppName);
+        Assert.Equal("3.6.5.0", s.AppVersion);
+        Assert.Equal("Succeeded", s.Status);
+        Assert.Equal("Installed", s.TerminalState);
+        Assert.Equal(0, s.ExitCode);
+        Assert.Equal(string.Empty, s.FailureCode);
+        Assert.Equal(T0, s.StartedAt);
+        Assert.Equal(T0, s.LastAttemptStartedAt);
+        Assert.Equal(T0.AddSeconds(22), s.CompletedAt);
+        Assert.Equal(22, s.DurationSeconds);
+        Assert.Equal(1, s.InstallPassCount);
+        Assert.True(MetricsMath.HasMeasuredDuration(s));
+    }
+
+    [Fact]
+    public void FailedPackage_ExitCodeIsTheFailureCode()
+    {
+        var s = Row(Aggregate(Started(T0), Completed(T0.AddSeconds(40), success: false, exitCode: 1603)), "generic-github-desktop");
+
+        Assert.Equal("Failed", s.Status);
+        Assert.Equal("Error", s.TerminalState);
+        Assert.Equal(1603, s.ExitCode);
+        Assert.Equal("1603", s.FailureCode);
+    }
+
+    [Fact]
+    public void StartedAndCompletedInOneRegistryPass_IsUnmeasured()
+    {
+        // Older agents report both events in the same watcher pass, milliseconds apart: that is
+        // no install duration.
+        var s = Row(Aggregate(Started(T0), Completed(T0.AddMilliseconds(4))), "generic-github-desktop");
+
+        Assert.Equal("Succeeded", s.Status);
+        Assert.Equal(0, s.DurationSeconds);
+        Assert.False(MetricsMath.HasMeasuredDuration(s));
+    }
+
+    [Fact]
+    public void MissingDisplayName_UsesPackageId_UntilANameArrives()
+    {
+        var summaries = Aggregate(Started(T0, displayName: string.Empty));
+        Assert.Equal("generic-github-desktop", Row(summaries, "generic-github-desktop").AppName);
+
+        EventIngestProcessor.AggregateAppInstallEvent(Completed(T0.AddSeconds(5)), "tenant", SessionId, summaries);
+        Assert.Equal("GitHub Desktop", Row(summaries, "generic-github-desktop").AppName);
+    }
+
+    [Fact]
+    public void SamePackageIdInBothScopes_IsTwoRows()
+    {
+        var summaries = Aggregate(
+            Completed(T0, scope: "machine"),
+            Completed(T0, success: false, exitCode: 1, scope: "user"));
+
+        Assert.Equal("Succeeded", Row(summaries, "generic-github-desktop", "machine").Status);
+        Assert.Equal("Failed", Row(summaries, "generic-github-desktop", "user").Status);
+    }
+
+    [Fact]
+    public void RealmJoinPackage_AndIntuneApp_WithTheSameName_AreTwoRows()
+    {
+        var ime = new EnrollmentEvent
+        {
+            EventType = "app_install_completed",
+            Timestamp = T0,
+            Data = new Dictionary<string, object> { ["appName"] = "GitHub Desktop", ["state"] = "Installed" },
+        };
+        var summaries = Aggregate(ime, Completed(T0.AddSeconds(1)));
+
+        Assert.Equal(2, summaries.Count);
+        Assert.Equal(string.Empty, summaries["GitHub Desktop"].Summary.Source);
+        Assert.Equal(AppInstallSources.RealmJoin, Row(summaries, "generic-github-desktop").Source);
+    }
+
+    [Fact]
+    public void EventWithoutPackageId_IsIgnored()
+    {
+        var evt = Completed(T0);
+        evt.Data!.Remove("packageId");
+        Assert.Empty(Aggregate(evt));
+    }
+
+    // ── store: row identity, channel columns, reconcile ─────────────────────
+
+    [Fact]
+    public void RowKey_ImeUnchanged_RealmJoinKeyedByScopeAndPackageId()
+    {
+        Assert.Equal($"{SessionId}_Contoso App_2",
+            TableStorageService.BuildAppInstallSummaryRowKey(new AppInstallSummary { SessionId = SessionId, AppName = "Contoso App/2" }));
+
+        var rj = new AppInstallSummary
+        {
+            SessionId = SessionId,
+            Source = AppInstallSources.RealmJoin,
+            InstallScope = "user",
+            AppId = "generic-7zip-usersettings",
+            AppName = "7-Zip Settings",
+        };
+        var key = TableStorageService.BuildAppInstallSummaryRowKey(rj);
+        Assert.Equal($"{SessionId}_realmjoin|user|generic-7zip-usersettings", key);
+        // The per-session range scans (ESP stamp, close step, deletion) match on this prefix.
+        Assert.StartsWith($"{SessionId}_", key, StringComparison.Ordinal);
+    }
+
+    [Fact]
+    public void Entity_ImeRowHasNoChannelColumns_RealmJoinRowRoundTrips()
+    {
+        var ime = TableStorageService.BuildAppInstallSummaryEntity(
+            new AppInstallSummary { TenantId = "t", SessionId = SessionId, AppName = "App" }, "rk");
+        Assert.False(ime.ContainsKey("Source"));
+        Assert.False(ime.ContainsKey("InstallScope"));
+
+        var summary = new AppInstallSummary
+        {
+            TenantId = "t",
+            SessionId = SessionId,
+            AppName = "7-Zip",
+            Source = AppInstallSources.RealmJoin,
+            InstallScope = "machine",
+            AppId = "generic-7zip",
+        };
+        var entity = TableStorageService.BuildAppInstallSummaryEntity(summary, "rk");
+        var sut = new TableStorageService(new Mock<TableServiceClient>().Object, NullLogger<TableStorageService>.Instance);
+        var mapped = sut.MapToAppInstallSummary(entity);
+
+        Assert.Equal(AppInstallSources.RealmJoin, mapped.Source);
+        Assert.Equal("machine", mapped.InstallScope);
+        Assert.Equal("generic-7zip", mapped.AppId);
+    }
+
+    [Fact]
+    public void Reconcile_PackageIdFallbackName_KeepsTheStoredDisplayName()
+    {
+        var existing = new TableEntity("t", "rk") { ["AppName"] = "7-Zip", ["StartedAt"] = T0 };
+        var summary = new AppInstallSummary
+        {
+            Source = AppInstallSources.RealmJoin,
+            AppId = "generic-7zip",
+            AppName = "generic-7zip",
+            StartedAt = T0.AddMinutes(30),
+        };
+
+        TableStorageService.ReconcileAppInstallSummaryWithExisting(summary, existing);
+        Assert.Equal("7-Zip", summary.AppName);
+    }
+
+    [Fact]
+    public void EspBlockingStamp_NeverTouchesRealmJoinRows()
+    {
+        var sets = EspBlockingSets.FromEventData(new Dictionary<string, object>
+        {
+            ["espTrackedWin32AppIds"] = new[] { "generic-7zip" },
+        })!;
+        var rj = new TableEntity("t", "rk") { ["Source"] = AppInstallSources.RealmJoin, ["AppId"] = "generic-7zip" };
+        var ime = new TableEntity("t", "rk2") { ["AppId"] = "generic-7zip" };
+
+        Assert.False(TableStorageService.ShouldStampEspBlocking(rj, sets));
+        Assert.True(TableStorageService.ShouldStampEspBlocking(ime, sets));
+    }
+
+    // ── per-app aggregates: the channel is part of the app ──────────────────
+
+    private static AppInstallSummary Summary(string appName, string source, string status = "Succeeded", int duration = 60)
+        => new()
+        {
+            TenantId = "t",
+            SessionId = Guid.NewGuid().ToString(),
+            AppName = appName,
+            Source = source,
+            Status = status,
+            TerminalState = status == "Failed" ? "Error" : "Installed",
+            DurationSeconds = duration,
+            StartedAt = DateTime.UtcNow.AddDays(-1),
+            LastAttemptStartedAt = DateTime.UtcNow.AddDays(-1),
+        };
+
+    private static List<AppInstallSummary> MixedChannelRows() => new()
+    {
+        Summary("GitHub Desktop", string.Empty),
+        Summary("GitHub Desktop", AppInstallSources.RealmJoin, status: "Failed"),
+        Summary("GitHub Desktop", AppInstallSources.RealmJoin),
+    };
+
+    [Fact]
+    public void AppsList_SameNameInTwoChannels_IsTwoItems()
+    {
+        var root = TestWire.SerializeToElement(AppsAnalyticsHelper.BuildAppsListResponse(MixedChannelRows(), days: 30));
+        var apps = root.GetProperty("apps").EnumerateArray().ToList();
+
+        Assert.Equal(2, apps.Count);
+        var ime = apps.Single(a => a.GetProperty("source").GetString() == AppInstallSources.Ime);
+        var rj = apps.Single(a => a.GetProperty("source").GetString() == AppInstallSources.RealmJoin);
+        Assert.Equal(1, ime.GetProperty("totalInstalls").GetInt32());
+        Assert.Equal(2, rj.GetProperty("totalInstalls").GetInt32());
+        Assert.Equal(1, rj.GetProperty("failed").GetInt32());
+    }
+
+    [Fact]
+    public async Task Analytics_And_Sessions_FilterByChannel()
+    {
+        var sessionRepo = new Mock<ISessionRepository>();
+        sessionRepo.Setup(r => r.GetSessionAsync(It.IsAny<string>(), It.IsAny<string>())).ReturnsAsync((SessionSummary?)null);
+
+        var rj = await AppsAnalyticsHelper.BuildAnalyticsResponseAsync(
+            MixedChannelRows(), sessionRepo.Object, "GitHub Desktop", AppInstallSources.RealmJoin, days: 30);
+        Assert.Equal(AppInstallSources.RealmJoin, rj.Source);
+        Assert.Equal(2, rj.Summary.TotalInstalls);
+        Assert.Equal(1, rj.Summary.Failed);
+
+        var ime = await AppsAnalyticsHelper.BuildAnalyticsResponseAsync(
+            MixedChannelRows(), sessionRepo.Object, "GitHub Desktop", AppInstallSources.Ime, days: 30);
+        Assert.Equal(AppInstallSources.Ime, ime.Source);
+        Assert.Equal(1, ime.Summary.TotalInstalls);
+
+        var sessions = await AppsAnalyticsHelper.BuildSessionsResponseAsync(
+            MixedChannelRows(), sessionRepo.Object, "GitHub Desktop", AppInstallSources.RealmJoin, days: 30,
+            statusFilter: "all", modelFilter: null, versionFilter: null, offset: 0, limit: 50);
+        Assert.Equal(2, sessions.Total);
+    }
+
+    [Fact]
+    public void AppMetrics_GroupsPerChannel()
+    {
+        var payload = MetricsMath.BuildAppMetricsPayload(MixedChannelRows());
+
+        Assert.Equal(2, payload.TotalApps);
+        var failing = Assert.Single(payload.TopFailingApps);
+        Assert.Equal(AppInstallSources.RealmJoin, failing.Source);
+        Assert.Equal("GitHub Desktop", failing.AppName);
+    }
+
+    [Theory]
+    [InlineData(null, true, "ime")]
+    [InlineData("", true, "ime")]
+    [InlineData("realmjoin", true, "realmjoin")]
+    [InlineData(" RealmJoin ", true, "realmjoin")]
+    [InlineData("office-c2r", false, "office-c2r")]
+    public void SourceQueryParam(string? raw, bool valid, string expected)
+    {
+        Assert.Equal(valid, AppsAnalyticsHelper.TryParseSourceQueryParam(raw, out var source));
+        Assert.Equal(expected, source);
+    }
+}

src/Backend/AutopilotMonitor.Functions.Tests/ReportAgentErrorFunctionTests.cs

@@ -4,9 +4,11 @@ using System.Net.Http;
 using System.Threading.Tasks;
 using AutopilotMonitor.Functions.Functions.Ingest;
 using AutopilotMonitor.Functions.Services;
+using AutopilotMonitor.Functions.Services.Deletion;
 using AutopilotMonitor.Functions.Services.Notifications;
 using AutopilotMonitor.Shared.DataAccess;
 using AutopilotMonitor.Shared.Models;
+using Azure.Data.Tables;
 using Microsoft.Extensions.Caching.Memory;
 using Microsoft.Extensions.Logging.Abstractions;
 using Moq;
@@ -94,14 +96,29 @@ public class ReportAgentErrorFunctionTests
     // (2026-07-23 hardening): timeline event + cross-session EventType index row +
     // AgentEmergencyBreak ops event, once per session. The index upsert is load-bearing:
     // StoreEventsBatchAsync alone leaves the event invisible to every search-by-eventType
-    // surface (found in the 2026-07-22 incident analysis).
+    // surface (found in the 2026-07-22 incident analysis). A session without a writable
+    // row gets the ops event only.
     // =========================================================================
 
+    private static TableEntity Row(SessionStatus? status = SessionStatus.InProgress, string? deletionState = null)
+    {
+        var row = new TableEntity(TenantId, SessionId);
+        if (status.HasValue) row["Status"] = status.Value.ToString();
+        if (deletionState != null) row["DeletionState"] = deletionState;
+        return row;
+    }
+
+    private static ReportAgentErrorFunction.SessionRowRead Read(TableEntity? row) => new(row, Failed: false);
+
+    private static readonly ReportAgentErrorFunction.SessionRowRead ReadFailed = new(null, Failed: true);
+
     private sealed class Harness
     {
         public readonly Mock<ISessionRepository> SessionRepo = new();
         public readonly List<OpsEventEntry> OpsEvents = new();
         public readonly OpsEventService OpsService;
+        private readonly SessionDeletionGuard _deletionGuard = new(
+            Mock.Of<ISessionDeletionInventoryReader>(), NullLogger<SessionDeletionGuard>.Instance);
 
         public Harness(List<EnrollmentEvent>? existingEvents = null)
         {
@@ -125,9 +142,23 @@ public class ReportAgentErrorFunctionTests
             OpsService = new OpsEventService(opsRepo.Object, NullLogger<OpsEventService>.Instance, alertDispatch);
         }
 
+        /// <summary>Default read: an existing, open session row — the historical path.</summary>
         public Task RunAsync(AgentErrorReport report)
+            => RunAsync(report, Read(Row()));
+
+        public Task RunAsync(AgentErrorReport report, ReportAgentErrorFunction.SessionRowRead? sessionRead)
             => ReportAgentErrorFunction.MaterializeEmergencyBreakArtifactsAsync(
-                report, TenantId, SessionRepo.Object, OpsService, NullLogger<ReportAgentErrorFunction>.Instance);
+                report, TenantId, sessionRead, SessionRepo.Object, _deletionGuard, OpsService,
+                NullLogger<ReportAgentErrorFunction>.Instance);
+
+        public void VerifyNothingWrittenIntoTheSession()
+        {
+            SessionRepo.Verify(r => r.GetSessionEventsAsync(
+                It.IsAny<string>(), It.IsAny<string>(), It.IsAny<int>()), Times.Never);
+            SessionRepo.Verify(r => r.StoreEventsBatchAsync(It.IsAny<List<EnrollmentEvent>>()), Times.Never);
+            SessionRepo.Verify(r => r.UpsertEventTypeIndexBatchAsync(
+                It.IsAny<string>(), It.IsAny<string>(), It.IsAny<IEnumerable<EnrollmentEvent>>()), Times.Never);
+        }
     }
 
     [Fact]
@@ -165,7 +196,7 @@ public class ReportAgentErrorFunctionTests
     }
 
     [Fact]
-    public async Task Non_break_error_types_and_missing_session_id_are_ignored()
+    public async Task Non_break_error_types_and_ids_without_a_row_read_are_ignored()
     {
         var h = new Harness();
 
@@ -173,12 +204,10 @@ public class ReportAgentErrorFunctionTests
         other.ErrorType = AgentErrorType.ConfigFetchFailed;
         await h.RunAsync(other);
 
-        var noSession = Report(Now);
-        noSession.SessionId = null!;
-        await h.RunAsync(noSession);
+        // No read = the ids were not GUIDs (ReadSessionRowAsync) — no session can exist to break.
+        await h.RunAsync(Report(Now), sessionRead: null);
 
-        h.SessionRepo.Verify(r => r.GetSessionEventsAsync(
-            It.IsAny<string>(), It.IsAny<string>(), It.IsAny<int>()), Times.Never);
+        h.VerifyNothingWrittenIntoTheSession();
         Assert.Empty(h.OpsEvents);
     }
 
@@ -192,10 +221,8 @@ public class ReportAgentErrorFunctionTests
     public async Task Break_on_already_terminal_session_is_info_late_cleanup(SessionStatus status)
     {
         var h = new Harness();
-        h.SessionRepo.Setup(r => r.GetSessionAsync(TenantId, SessionId))
-            .ReturnsAsync(new SessionSummary { SessionId = SessionId, TenantId = TenantId, Status = status });
 
-        await h.RunAsync(Report(Now));
+        await h.RunAsync(Report(Now), Read(Row(status)));
 
         var ops = Assert.Single(h.OpsEvents);
         Assert.Equal(OpsEventSeverity.Info, ops.Severity);
@@ -216,44 +243,220 @@ public class ReportAgentErrorFunctionTests
     public async Task Break_on_open_session_stays_warning(SessionStatus status)
     {
         var h = new Harness();
-        h.SessionRepo.Setup(r => r.GetSessionAsync(TenantId, SessionId))
-            .ReturnsAsync(new SessionSummary { SessionId = SessionId, TenantId = TenantId, Status = status });
 
-        await h.RunAsync(Report(Now));
+        await h.RunAsync(Report(Now), Read(Row(status)));
 
         var ops = Assert.Single(h.OpsEvents);
         Assert.Equal(OpsEventSeverity.Warning, ops.Severity);
         Assert.Contains($"still {status}", ops.Message);
         Assert.Contains("\"lateCleanup\":false", ops.Details);
+        h.SessionRepo.Verify(r => r.StoreEventsBatchAsync(It.IsAny<List<EnrollmentEvent>>()), Times.Once);
+    }
+
+    // 2026-09-28 sweep: every "unknown" Warning named a session that never registered. The break is
+    // the only thing the backend ever hears of it — nothing to close, and a timeline event would be
+    // an orphan the maintenance sweep deletes a day later (with a Warning of its own).
+    [Fact]
+    public async Task Break_on_session_without_row_is_info_and_writes_nothing_into_the_session()
+    {
+        var h = new Harness();
+
+        await h.RunAsync(Report(Now), Read(null));
+
+        var ops = Assert.Single(h.OpsEvents);
+        Assert.Equal(OpsEventSeverity.Info, ops.Severity);
+        Assert.Contains("no session row", ops.Message);
+        Assert.Contains("\"sessionStatusAtBreak\":\"missing\"", ops.Details);
+        Assert.Contains("\"lateCleanup\":false", ops.Details);
+        h.VerifyNothingWrittenIntoTheSession();
+    }
+
+    [Theory]
+    [InlineData(AutopilotMonitor.Shared.Models.Deletion.SessionDeletionState.Preparing)]
+    [InlineData(AutopilotMonitor.Shared.Models.Deletion.SessionDeletionState.Queued)]
+    [InlineData(AutopilotMonitor.Shared.Models.Deletion.SessionDeletionState.Running)]
+    [InlineData(AutopilotMonitor.Shared.Models.Deletion.SessionDeletionState.Poisoned)]
+    public async Task Break_on_session_under_cascade_delete_writes_nothing_past_the_lock(string deletionState)
+    {
+        var h = new Harness();
+
+        await h.RunAsync(Report(Now), Read(Row(SessionStatus.InProgress, deletionState)));
+
+        var ops = Assert.Single(h.OpsEvents);
+        Assert.Equal(OpsEventSeverity.Info, ops.Severity);
+        Assert.Contains("\"sessionStatusAtBreak\":\"deleting\"", ops.Details);
+        h.VerifyNothingWrittenIntoTheSession();
     }
 
     [Fact]
-    public async Task Unknown_or_unreadable_session_status_keeps_warning()
+    public async Task Unreadable_row_keeps_the_historical_warning_and_write()
     {
-        // No row (Harness default: GetSessionAsync → null) — the historical Warning semantics.
+        // Existence unknown — fail safe: the break may be the only signal for an open session.
         var h = new Harness();
-        await h.RunAsync(Report(Now));
+
+        await h.RunAsync(Report(Now), ReadFailed);
+
         var ops = Assert.Single(h.OpsEvents);
         Assert.Equal(OpsEventSeverity.Warning, ops.Severity);
         Assert.Contains("\"sessionStatusAtBreak\":\"unknown\"", ops.Details);
-
-        // Row read throws — the break must still be recorded, still as Warning.
-        var h2 = new Harness();
-        h2.SessionRepo.Setup(r => r.GetSessionAsync(TenantId, SessionId))
-            .ThrowsAsync(new InvalidOperationException("table down"));
-        await h2.RunAsync(Report(Now));
-        Assert.Equal(OpsEventSeverity.Warning, Assert.Single(h2.OpsEvents).Severity);
+        h.SessionRepo.Verify(r => r.StoreEventsBatchAsync(It.IsAny<List<EnrollmentEvent>>()), Times.Once);
     }
 
     [Fact]
-    public void ClassifyEmergencyBreak_is_terminal_set_only()
+    public async Task Row_without_parseable_status_stays_warning()
     {
-        Assert.Equal((OpsEventSeverity.Info, true), ReportAgentErrorFunction.ClassifyEmergencyBreak(SessionStatus.Succeeded));
-        Assert.Equal((OpsEventSeverity.Info, true), ReportAgentErrorFunction.ClassifyEmergencyBreak(SessionStatus.Failed));
-        Assert.Equal((OpsEventSeverity.Info, true), ReportAgentErrorFunction.ClassifyEmergencyBreak(SessionStatus.Incomplete));
-        Assert.Equal((OpsEventSeverity.Warning, false), ReportAgentErrorFunction.ClassifyEmergencyBreak(SessionStatus.AwaitingUser));
-        Assert.Equal((OpsEventSeverity.Warning, false), ReportAgentErrorFunction.ClassifyEmergencyBreak(SessionStatus.Unknown));
-        Assert.Equal((OpsEventSeverity.Warning, false), ReportAgentErrorFunction.ClassifyEmergencyBreak(null));
+        var h = new Harness();
+
+        await h.RunAsync(Report(Now), Read(Row(status: null)));
+
+        var ops = Assert.Single(h.OpsEvents);
+        Assert.Equal(OpsEventSeverity.Warning, ops.Severity);
+        Assert.Contains("\"sessionStatusAtBreak\":\"unknown\"", ops.Details);
+    }
+
+    // The agent-measured age separates a break just past the cap from a device that slept for
+    // weeks; the backend cannot derive it (no row, or a row registered seconds before the break).
+    [Fact]
+    public async Task Reported_session_age_lands_in_ops_details_and_timeline_data()
+    {
+        var h = new Harness();
+        var report = Report(Now);
+        report.SessionAgeHours = 478.2;
+
+        await h.RunAsync(report);
+
+        Assert.Contains("\"sessionAgeHours\":478.2", Assert.Single(h.OpsEvents).Details);
+        h.SessionRepo.Verify(r => r.StoreEventsBatchAsync(
+            It.Is<List<EnrollmentEvent>>(e => (double)e[0].Data["sessionAgeHours"] == 478.2)), Times.Once);
+    }
+
+    // A session that never registered leaves the backend nothing but this report; the agent's
+    // record of its failed starts is the only account of why.
+    [Fact]
+    public async Task Prior_registration_failure_lands_in_ops_details_even_without_a_row()
+    {
+        var h = new Harness();
+        var report = Report(Now);
+        report.PriorRegistrationFailure = ReportAgentErrorFunction.SanitizeRegistrationFailure(Failure());
+
+        await h.RunAsync(report, Read(null));
+
+        var details = Assert.Single(h.OpsEvents).Details;
+        Assert.Contains("\"priorRegistrationFailure\":{\"failedRuns\":3", details);
+        Assert.Contains("\"networkLinkUpAtEnd\":false", details);
+        Assert.Contains("\"attemptWindowSeconds\":1204.5", details);
+        Assert.Contains("\"configFetchOutcome\":\"UsedDefaults\"", details);
+        h.VerifyNothingWrittenIntoTheSession();
+    }
+
+    [Fact]
+    public async Task Prior_registration_failure_lands_in_the_timeline_data_of_a_registered_session()
+    {
+        var h = new Harness();
+        var report = Report(Now);
+        report.PriorRegistrationFailure = ReportAgentErrorFunction.SanitizeRegistrationFailure(Failure());
+
+        await h.RunAsync(report);
+
+        h.SessionRepo.Verify(r => r.StoreEventsBatchAsync(
+            It.Is<List<EnrollmentEvent>>(e =>
+                ((Dictionary<string, object?>)e[0].Data["priorRegistrationFailure"])["failedRuns"]!.Equals(3))), Times.Once);
+    }
+
+    [Fact]
+    public async Task Agent_without_registration_record_leaves_it_null_and_out_of_the_timeline()
+    {
+        var h = new Harness();
+
+        await h.RunAsync(Report(Now));
+
+        Assert.Contains("\"priorRegistrationFailure\":null", Assert.Single(h.OpsEvents).Details);
+        h.SessionRepo.Verify(r => r.StoreEventsBatchAsync(
+            It.Is<List<EnrollmentEvent>>(e => !e[0].Data.ContainsKey("priorRegistrationFailure"))), Times.Once);
+    }
+
+    [Fact]
+    public void SanitizeRegistrationFailure_bounds_device_input()
+    {
+        var raw = Failure();
+        raw.LastError = new string('x', 5000);
+        raw.Outcome = new string('o', 500);
+        raw.ConfigFetchOutcome = new string('c', 500);
+        raw.AttemptWindowSeconds = -3;
+
+        var clean = ReportAgentErrorFunction.SanitizeRegistrationFailure(raw)!;
+
+        Assert.Equal(ReportAgentErrorFunction.MaxRegistrationErrorLength, clean.LastError!.Length);
+        Assert.Equal(ReportAgentErrorFunction.MaxRegistrationTokenLength, clean.Outcome.Length);
+        Assert.Equal(ReportAgentErrorFunction.MaxRegistrationTokenLength, clean.ConfigFetchOutcome!.Length);
+        Assert.Equal(0, clean.AttemptWindowSeconds);
+        Assert.Equal(3, clean.FailedRuns);
+        Assert.NotSame(raw, clean);
+    }
+
+    [Theory]
+    [InlineData(0)]
+    [InlineData(-1)]
+    public void SanitizeRegistrationFailure_drops_a_record_without_failed_runs(int failedRuns)
+    {
+        var raw = Failure();
+        raw.FailedRuns = failedRuns;
+
+        Assert.Null(ReportAgentErrorFunction.SanitizeRegistrationFailure(raw));
+        Assert.Null(ReportAgentErrorFunction.SanitizeRegistrationFailure(null));
+    }
+
+    private static RegistrationFailureSummary Failure() => new()
+    {
+        FailedRuns = 3,
+        FirstFailedAtUtc = Now.AddDays(-20),
+        LastFailedAtUtc = Now.AddDays(-19),
+        Outcome = "Failed",
+        LastError = "No such host is known.",
+        NetworkLinkUpAtStart = false,
+        NetworkLinkUpAtEnd = false,
+        AttemptWindowSeconds = 1204.5,
+        ConfigFetchOutcome = "UsedDefaults",
+    };
+
+    [Fact]
+    public async Task Agent_without_session_age_leaves_it_null_and_out_of_the_timeline()
+    {
+        var h = new Harness();
+
+        await h.RunAsync(Report(Now));
+
+        Assert.Contains("\"sessionAgeHours\":null", Assert.Single(h.OpsEvents).Details);
+        h.SessionRepo.Verify(r => r.StoreEventsBatchAsync(
+            It.Is<List<EnrollmentEvent>>(e => !e[0].Data.ContainsKey("sessionAgeHours"))), Times.Once);
+    }
+
+    [Fact]
+    public void ClassifyEmergencyBreak_follows_row_state()
+    {
+        static ReportAgentErrorFunction.EmergencyBreakVerdict Classify(TableEntity? row, bool failed = false, bool locked = false)
+            => ReportAgentErrorFunction.ClassifyEmergencyBreak(row, failed, locked);
+
+        foreach (var terminal in new[] { SessionStatus.Succeeded, SessionStatus.Failed, SessionStatus.Incomplete })
+        {
+            var v = Classify(Row(terminal));
+            Assert.Equal((OpsEventSeverity.Info, true, true, terminal.ToString()), (v.Severity, v.LateCleanup, v.Materialize, v.SessionStatusAtBreak));
+        }
+
+        foreach (var open in new[] { SessionStatus.InProgress, SessionStatus.Pending, SessionStatus.AwaitingUser, SessionStatus.Stalled, SessionStatus.Unknown })
+        {
+            var v = Classify(Row(open));
+            Assert.Equal((OpsEventSeverity.Warning, false, true, open.ToString()), (v.Severity, v.LateCleanup, v.Materialize, v.SessionStatusAtBreak));
+        }
+
+        var missing = Classify(null);
+        Assert.Equal((OpsEventSeverity.Info, false, false, "missing"), (missing.Severity, missing.LateCleanup, missing.Materialize, missing.SessionStatusAtBreak));
+
+        var deleting = Classify(Row(SessionStatus.Succeeded), locked: true);
+        Assert.Equal((OpsEventSeverity.Info, false, false, "deleting"), (deleting.Severity, deleting.LateCleanup, deleting.Materialize, deleting.SessionStatusAtBreak));
+
+        var unreadable = Classify(null, failed: true);
+        Assert.Equal((OpsEventSeverity.Warning, false, true, "unknown"), (unreadable.Severity, unreadable.LateCleanup, unreadable.Materialize, unreadable.SessionStatusAtBreak));
     }
 
     [Fact]

src/Backend/AutopilotMonitor.Functions.Tests/RevokeEnforcementTests.cs

@@ -373,4 +373,106 @@ public class RevokeEnforcementTests
 
         proConferral.Verify(p => p.RecordLossAsync(It.IsAny<string>(), It.IsAny<string>(), It.IsAny<string>()), Times.Never);
     }
+
+    // --- Customer-visible audit rows name the group as the customer sees it, never internally ---
+
+    [Theory]
+    [InlineData(null, "Platform support")]
+    [InlineData("Contoso Service Desk", "Contoso Service Desk")]
+    public async Task UnassignCore_AuditNamesTheGroupByItsCustomerName(string? label, string expected)
+    {
+        var (fn, repo, audit, _) = BuildGroupFn();
+        repo.Setup(r => r.GetTenantGroupAsync(GroupId)).ReturnsAsync(new TenantGroup
+        {
+            GroupId = GroupId,
+            Name = "Internal MSP customers north",
+            CustomerLabel = label,
+            TenantIds = new List<string> { TenantA },
+            AssigneeCount = 1,
+        });
+        repo.Setup(r => r.GetGroupAssignmentsForUpnAsync(UpnLower))
+            .ReturnsAsync(new List<TenantGroupAssignment> { new() { Upn = UpnLower, GroupId = GroupId } });
+        repo.Setup(r => r.UnassignGroupAsync(UpnLower, GroupId)).ReturnsAsync(true);
+
+        await fn.UnassignCoreAsync(GroupId, Upn, "ga@vendor.example");
+
+        audit.Verify(a => a.LogAuditEntryAsync(
+            TenantA, "DELETE", "DelegatedGroupAccess", UpnLower, "ga@vendor.example",
+            It.Is<Dictionary<string, string>?>(d => d != null && d["Group"] == expected && d["GroupId"] == GroupId)), Times.Once);
+    }
+
+    // --- Group update: name and/or customer label (PATCH only what changes) ---
+
+    [Theory]
+    [InlineData(null, null, "name or customerLabel is required")]
+    [InlineData("  ", null, "name must not be empty")]
+    [InlineData("Renamed", null, null)]
+    [InlineData(null, "", null)] // clears the label
+    [InlineData(null, "Contoso Service Desk", null)]
+    public void ValidateUpdateRequest_Shapes(string? name, string? label, string? expectedError)
+        => Assert.Equal(expectedError,
+            TenantGroupManagementFunction.ValidateUpdateRequest(new UpdateTenantGroupRequest { Name = name, CustomerLabel = label }));
+
+    [Fact]
+    public void ValidateUpdateRequest_LabelLongerThanCap_Rejected_TrimmedLengthCounts()
+    {
+        var atCap = "  " + new string('x', DelegationSelfService.CustomerLabelMaxLength) + "  ";
+        Assert.Null(TenantGroupManagementFunction.ValidateUpdateRequest(new UpdateTenantGroupRequest { CustomerLabel = atCap }));
+
+        var tooLong = new string('x', DelegationSelfService.CustomerLabelMaxLength + 1);
+        Assert.NotNull(TenantGroupManagementFunction.ValidateUpdateRequest(new UpdateTenantGroupRequest { CustomerLabel = tooLong }));
+    }
+
+    [Fact]
+    public async Task UpdateGroupCore_Label_OnOperatorGroup_WritesOnlyTheLabel_NoAudit()
+    {
+        var (fn, repo, audit, _) = BuildGroupFn();
+        repo.Setup(r => r.GetTenantGroupAsync(GroupId)).ReturnsAsync(new TenantGroup { GroupId = GroupId, Name = "MSP Customers" });
+        repo.Setup(r => r.UpdateTenantGroupAsync(GroupId, null, "Contoso Service Desk")).ReturnsAsync(true);
+
+        var outcome = await fn.UpdateGroupCoreAsync(GroupId, new UpdateTenantGroupRequest { CustomerLabel = "Contoso Service Desk" }, "ga@vendor.example");
+
+        Assert.Equal(TenantGroupManagementFunction.GroupUpdateOutcome.Updated, outcome);
+        repo.Verify(r => r.UpdateTenantGroupAsync(GroupId, null, "Contoso Service Desk"), Times.Once);
+        VerifyNoAudit(audit); // naming is not an access change
+    }
+
+    [Fact]
+    public async Task UpdateGroupCore_Label_OnOwnedGroup_Refused_NothingWritten()
+    {
+        var (fn, repo, _, _) = BuildGroupFn();
+        var ownedGroupId = Constants.TenantGroupIds.ForHomeTenant(OwnerTenant);
+        repo.Setup(r => r.GetTenantGroupAsync(ownedGroupId)).ReturnsAsync(new TenantGroup { GroupId = ownedGroupId, Name = "partner.example", OwnerTenantId = OwnerTenant });
+
+        var outcome = await fn.UpdateGroupCoreAsync(ownedGroupId, new UpdateTenantGroupRequest { CustomerLabel = "Anything" }, "ga@vendor.example");
+
+        Assert.Equal(TenantGroupManagementFunction.GroupUpdateOutcome.LabelOnOwnedGroup, outcome);
+        repo.Verify(r => r.UpdateTenantGroupAsync(It.IsAny<string>(), It.IsAny<string?>(), It.IsAny<string?>()), Times.Never);
+    }
+
+    [Fact]
+    public async Task UpdateGroupCore_Label_UnknownGroup_NotFound()
+    {
+        var (fn, repo, _, _) = BuildGroupFn(); // default: no group
+
+        var outcome = await fn.UpdateGroupCoreAsync(GroupId, new UpdateTenantGroupRequest { CustomerLabel = "Anything" }, "ga@vendor.example");
+
+        Assert.Equal(TenantGroupManagementFunction.GroupUpdateOutcome.NotFound, outcome);
+        repo.Verify(r => r.UpdateTenantGroupAsync(It.IsAny<string>(), It.IsAny<string?>(), It.IsAny<string?>()), Times.Never);
+    }
+
+    [Fact]
+    public async Task UpdateGroupCore_RenameOnly_LeavesTheLabelAlone_MissIsNotFound()
+    {
+        var (fn, repo, _, _) = BuildGroupFn();
+        repo.Setup(r => r.UpdateTenantGroupAsync(GroupId, "Renamed", null)).ReturnsAsync(true);
+
+        Assert.Equal(TenantGroupManagementFunction.GroupUpdateOutcome.Updated,
+            await fn.UpdateGroupCoreAsync(GroupId, new UpdateTenantGroupRequest { Name = "Renamed" }, "ga@vendor.example"));
+        repo.Verify(r => r.UpdateTenantGroupAsync(GroupId, "Renamed", null), Times.Once);
+
+        repo.Setup(r => r.UpdateTenantGroupAsync("tpl-missing", "Renamed", null)).ReturnsAsync(false);
+        Assert.Equal(TenantGroupManagementFunction.GroupUpdateOutcome.NotFound,
+            await fn.UpdateGroupCoreAsync("tpl-missing", new UpdateTenantGroupRequest { Name = "Renamed" }, "ga@vendor.example"));
+    }
 }

src/Backend/AutopilotMonitor.Functions.Tests/SessionOwnerBindingCallSiteGuardTests.cs

@@ -0,0 +1,63 @@
+using System;
+using System.Collections.Generic;
+using System.IO;
+using System.Linq;
+using Xunit;
+
+namespace AutopilotMonitor.Functions.Tests;
+
+/// <summary>
+/// SESSION-OWNER-BINDING call-site inventory. The functions are too dependency-heavy to boot in a
+/// unit test, and the refusal is a one-line branch at each site — so the guard is structural:
+/// every file that observes the binding must also act on <c>Decision.Rejected</c>, and the set of
+/// sites is pinned so a new session-scoped agent endpoint is added here on purpose (and decides
+/// how it refuses) instead of silently observing without enforcing.
+/// </summary>
+public class SessionOwnerBindingCallSiteGuardTests
+{
+    private const string ObserveCall = "_ownerBinding.Observe(";
+
+    private static readonly string[] ExpectedSites =
+    {
+        "Functions/Diagnostics/GetDiagnosticsUploadUrlFunction.cs",
+        "Functions/Ingest/IngestTelemetryFunction.cs",
+        "Functions/Ingest/ReportAgentErrorFunction.cs",
+        "Functions/Sessions/RegisterSessionFunction.cs",
+    };
+
+    [Fact]
+    public void Every_observing_site_acts_on_the_refusal()
+    {
+        var functionsRoot = Path.Combine(FindRepoRoot(), "src", "Backend", "AutopilotMonitor.Functions");
+        var sites = new Dictionary<string, string>(StringComparer.Ordinal);
+
+        foreach (var file in Directory.EnumerateFiles(functionsRoot, "*.cs", SearchOption.AllDirectories))
+        {
+            var relative = Path.GetRelativePath(functionsRoot, file).Replace(Path.DirectorySeparatorChar, '/');
+            if (relative.StartsWith("bin/", StringComparison.OrdinalIgnoreCase) ||
+                relative.StartsWith("obj/", StringComparison.OrdinalIgnoreCase))
+                continue;
+
+            var text = File.ReadAllText(file);
+            if (text.Contains(ObserveCall, StringComparison.Ordinal))
+                sites[relative] = text;
+        }
+
+        Assert.Equal(ExpectedSites, sites.Keys.OrderBy(k => k, StringComparer.Ordinal).ToArray());
+
+        var missing = sites.Where(kv => !kv.Value.Contains(".Rejected", StringComparison.Ordinal)).Select(kv => kv.Key).ToList();
+        Assert.True(missing.Count == 0,
+            "These sites observe the session owner binding but never act on Decision.Rejected:\n  " + string.Join("\n  ", missing));
+    }
+
+    private static string FindRepoRoot()
+    {
+        var dir = new DirectoryInfo(AppContext.BaseDirectory);
+        while (dir != null && !File.Exists(Path.Combine(dir.FullName, "AutopilotMonitor.sln")))
+        {
+            dir = dir.Parent;
+        }
+        Assert.NotNull(dir);
+        return dir!.FullName;
+    }
+}

src/Backend/AutopilotMonitor.Functions.Tests/SessionOwnerBindingObserverTests.cs

@@ -19,9 +19,9 @@ using Xunit;
 namespace AutopilotMonitor.Functions.Tests;
 
 /// <summary>
-/// SESSION-OWNER-BINDING-SHADOW carriers: the outcome lands on the request row for every
-/// request (denominator), non-Match outcomes produce a Warning, would-reject outcomes raise one
-/// throttled ops event, stamping is fail-soft, and nothing is ever refused.
+/// SESSION-OWNER-BINDING carriers: the outcome lands on the request row for every request
+/// (denominator), non-Match outcomes produce a Warning, rejected outcomes raise one throttled
+/// ops event, stamping is fail-soft, and an evaluation defect fails open.
 /// </summary>
 public class SessionOwnerBindingObserverTests
 {
@@ -105,7 +105,7 @@ public class SessionOwnerBindingObserverTests
     }
 
     [Fact]
-    public async Task Would_reject_outcome_warns_and_raises_one_ops_event_per_session_and_outcome()
+    public async Task Rejected_outcome_warns_and_raises_one_ops_event_per_session_and_outcome()
     {
         var rig = new Rig();
         var (req, items) = Request("2.0.1500");
@@ -115,12 +115,12 @@ public class SessionOwnerBindingObserverTests
         var second = rig.Sut.Observe(req, Tenant, Session, row, Cert(Thumb2, "SN-ATTACKER"), "agent/telemetry");
 
         Assert.Equal(SessionOwnershipPolicy.Outcome.MismatchCert, first.Outcome);
-        Assert.True(first.WouldReject);
+        Assert.True(first.Rejected);
         Assert.Equal(SessionOwnershipPolicy.Outcome.MismatchCert, items[SessionOwnershipPolicy.RequestItemKey]);
 
         var warnings = rig.Logger.Entries.Where(e => e.Level == LogLevel.Warning).ToList();
         Assert.Equal(2, warnings.Count);
-        Assert.All(warnings, w => Assert.Contains("AgentSessionOwnerBinding outcome=MismatchCert enforced=False wouldReject=True", w.Message));
+        Assert.All(warnings, w => Assert.Contains("AgentSessionOwnerBinding outcome=MismatchCert enforced=True wouldReject=True", w.Message));
         Assert.Contains("serialMatch=False", warnings[0].Message);
         Assert.Contains("ver=2.0.1500", warnings[0].Message);
 
@@ -130,11 +130,11 @@ public class SessionOwnerBindingObserverTests
         Assert.Equal(OpsEventCategory.Security, evt.Category);
         Assert.Equal(Tenant, evt.TenantId);
         Assert.Contains("MismatchCert", evt.Message);
-        Assert.Contains("shadow", evt.Message);
+        Assert.Contains("refused", evt.Message);
         Assert.DoesNotContain(Thumb1, evt.Message);
         Assert.DoesNotContain(Thumb2, evt.Message);
         Assert.Contains("\"serialMatch\":false", evt.Details);
-        Assert.Contains("\"enforced\":false", evt.Details);
+        Assert.Contains("\"enforced\":true", evt.Details);
         _ = second;
     }
 
@@ -154,6 +154,22 @@ public class SessionOwnerBindingObserverTests
         Assert.Empty(rig.Ops);
     }
 
+    [Fact]
+    public void Evaluation_defect_fails_open()
+    {
+        var rig = new Rig();
+        var (req, _) = Request();
+
+        // A null validation throws inside the policy — our defect, never evidence of a foreign device.
+        var d = rig.Sut.Observe(req, Tenant, Session, OwnedRow(Thumb1), null!, "agent/telemetry");
+
+        Assert.Equal(SessionOwnershipPolicy.Outcome.Match, d.Outcome);
+        Assert.False(d.Rejected);
+        Assert.Null(d.OwnerToStamp);
+        Assert.Contains(rig.Logger.Entries, e => e.Level == LogLevel.Warning && e.Message.Contains("request allowed"));
+        Assert.Empty(rig.Ops);
+    }
+
     [Fact]
     public async Task Stamp_writes_the_owner_and_swallows_storage_failures()
     {

src/Backend/AutopilotMonitor.Functions.Tests/SessionOwnershipPolicyTests.cs

@@ -11,8 +11,8 @@ namespace AutopilotMonitor.Functions.Tests;
 /// <summary>
 /// SESSION-OWNER-BINDING rule matrix. The policy is pure, so every lifecycle shape that must keep
 /// working (restart, WhiteGlove Part 2, bootstrap→cert handoff, certificate re-issue, rows from
-/// before the binding) and every shape that must be visible as a would-reject (foreign cert,
-/// foreign bootstrap code, bootstrap on a cert-owned row, re-enroll without wipe) is pinned here.
+/// before the binding) and every shape that is refused (foreign cert, foreign bootstrap code,
+/// bootstrap on a cert-owned row, re-enroll without wipe) is pinned here.
 /// </summary>
 public class SessionOwnershipPolicyTests
 {
@@ -65,7 +65,7 @@ public class SessionOwnershipPolicyTests
         Assert.Equal(Thumb1, d.OwnerToStamp.Thumbprint);
         Assert.Equal(Dev1, d.OwnerToStamp.DeviceId);
         Assert.Equal("SN-1", d.OwnerToStamp.Serial);
-        Assert.False(d.WouldReject);
+        Assert.False(d.Rejected);
     }
 
     [Fact]
@@ -75,7 +75,7 @@ public class SessionOwnershipPolicyTests
         var d = SessionOwnershipPolicy.Evaluate(Row(owner: CertOwner(Thumb1, Dev1)), v, Now);
         Assert.Equal(SessionOwnershipPolicy.Outcome.CallerUnidentified, d.Outcome);
         Assert.Null(d.OwnerToStamp);
-        Assert.False(d.WouldReject);
+        Assert.False(d.Rejected);
     }
 
     // ── legacy rows (pre-binding) ────────────────────────────────────────────
@@ -87,7 +87,7 @@ public class SessionOwnershipPolicyTests
         Assert.Equal(SessionOwnershipPolicy.Outcome.ClaimLegacy, d.Outcome);
         Assert.NotNull(d.OwnerToStamp);
         Assert.True(d.SerialMatch);
-        Assert.False(d.WouldReject);
+        Assert.False(d.Rejected);
     }
 
     [Fact]
@@ -96,7 +96,7 @@ public class SessionOwnershipPolicyTests
         var d = SessionOwnershipPolicy.Evaluate(Row(serial: "OTHER"), Cert(Thumb1, Dev1), Now);
         Assert.Equal(SessionOwnershipPolicy.Outcome.LegacySerialMismatch, d.Outcome);
         Assert.Null(d.OwnerToStamp);
-        Assert.True(d.WouldReject);
+        Assert.True(d.Rejected);
     }
 
     [Fact]
@@ -123,7 +123,7 @@ public class SessionOwnershipPolicyTests
         Assert.Equal(SessionOwnershipPolicy.Outcome.RebindCertRotation, d.Outcome);
         Assert.Equal(Thumb2, d.OwnerToStamp!.Thumbprint);
         Assert.Equal(Dev1, d.OwnerToStamp.DeviceId);
-        Assert.False(d.WouldReject);
+        Assert.False(d.Rejected);
     }
 
     [Fact]
@@ -133,7 +133,7 @@ public class SessionOwnershipPolicyTests
         Assert.Equal(SessionOwnershipPolicy.Outcome.MismatchCert, d.Outcome);
         Assert.False(d.SerialMatch);
         Assert.Null(d.OwnerToStamp);
-        Assert.True(d.WouldReject);
+        Assert.True(d.Rejected);
     }
 
     [Fact]
@@ -143,7 +143,7 @@ public class SessionOwnershipPolicyTests
         var d = SessionOwnershipPolicy.Evaluate(Row(owner: CertOwner(Thumb1, Dev1)), Cert(Thumb2, Dev2, "SN-1"), Now);
         Assert.Equal(SessionOwnershipPolicy.Outcome.MismatchCert, d.Outcome);
         Assert.True(d.SerialMatch);
-        Assert.True(d.WouldReject);
+        Assert.True(d.Rejected);
     }
 
     [Fact]
@@ -160,7 +160,7 @@ public class SessionOwnershipPolicyTests
         Assert.Equal(SessionOwnershipPolicy.Outcome.RebindBootstrapHandoff, d.Outcome);
         Assert.Equal(SessionOwner.Kinds.Cert, d.OwnerToStamp!.Kind);
         Assert.Null(d.OwnerToStamp.BootstrapCode);
-        Assert.False(d.WouldReject);
+        Assert.False(d.Rejected);
     }
 
     [Fact]
@@ -168,7 +168,7 @@ public class SessionOwnershipPolicyTests
     {
         var d = SessionOwnershipPolicy.Evaluate(Row(owner: BootstrapOwner("ABC123")), Cert(Thumb1, Dev1, "OTHER"), Now);
         Assert.Equal(SessionOwnershipPolicy.Outcome.MismatchBootstrapOwned, d.Outcome);
-        Assert.True(d.WouldReject);
+        Assert.True(d.Rejected);
     }
 
     // ── bootstrap caller ─────────────────────────────────────────────────────
@@ -187,7 +187,7 @@ public class SessionOwnershipPolicyTests
     {
         var d = SessionOwnershipPolicy.Evaluate(Row(owner: BootstrapOwner("ABC123")), Bootstrap(code, serial), Now);
         Assert.Equal(SessionOwnershipPolicy.Outcome.MismatchBootstrap, d.Outcome);
-        Assert.True(d.WouldReject);
+        Assert.True(d.Rejected);
     }
 
     [Fact]
@@ -195,7 +195,7 @@ public class SessionOwnershipPolicyTests
     {
         var d = SessionOwnershipPolicy.Evaluate(Row(owner: CertOwner(Thumb1, Dev1)), Bootstrap("ABC123"), Now);
         Assert.Equal(SessionOwnershipPolicy.Outcome.DowngradeToBootstrap, d.Outcome);
-        Assert.True(d.WouldReject);
+        Assert.True(d.Rejected);
     }
 
     [Fact]
@@ -244,7 +244,7 @@ public class SessionOwnershipPolicyTests
     [InlineData(SessionOwnershipPolicy.Outcome.RebindBootstrapHandoff)]
     [InlineData(SessionOwnershipPolicy.Outcome.CallerUnidentified)]
     public void Lifecycle_outcomes_are_deliberately_tolerated(string outcome)
-        => Assert.False(SessionOwnershipPolicy.WouldRejectUnderEnforcement(outcome));
+        => Assert.False(SessionOwnershipPolicy.Rejects(outcome));
 
     [Theory]
     [InlineData(SessionOwnershipPolicy.Outcome.LegacySerialMismatch)]
@@ -252,16 +252,22 @@ public class SessionOwnershipPolicyTests
     [InlineData(SessionOwnershipPolicy.Outcome.MismatchCert)]
     [InlineData(SessionOwnershipPolicy.Outcome.MismatchBootstrap)]
     [InlineData(SessionOwnershipPolicy.Outcome.DowngradeToBootstrap)]
-    public void Foreign_identity_outcomes_would_reject(string outcome)
-        => Assert.True(SessionOwnershipPolicy.WouldRejectUnderEnforcement(outcome));
+    public void Foreign_identity_outcomes_are_rejected(string outcome)
+        => Assert.True(SessionOwnershipPolicy.Rejects(outcome));
 
     [Fact]
-    public void Stage1_has_no_Rejects_rule()
+    public void Every_outcome_is_classified_and_the_telemetry_field_reports_the_rule()
     {
-        // Enforcement is a deliberate, visible code change (stage 2) — not something that can
-        // slip in through a refactor. Delete this test in the change that adds Rejects.
-        var rejects = typeof(SessionOwnershipPolicy).GetMethods(BindingFlags.Public | BindingFlags.Static)
-            .Where(m => m.Name == "Rejects").ToArray();
-        Assert.Empty(rejects);
+        // A new Outcome constant must land in one of the two theories above on purpose, and the
+        // shadow-era wouldReject field (KQL keeps querying it) must say exactly what Rejects does.
+        var outcomes = typeof(SessionOwnershipPolicy.Outcome)
+            .GetFields(BindingFlags.Public | BindingFlags.Static)
+            .Where(f => f.IsLiteral)
+            .Select(f => (string)f.GetRawConstantValue()!)
+            .ToArray();
+
+        Assert.Equal(11, outcomes.Length);
+        Assert.Equal(5, outcomes.Count(SessionOwnershipPolicy.Rejects));
+        Assert.All(outcomes, o => Assert.Equal(SessionOwnershipPolicy.Rejects(o), SessionOwnershipPolicy.WouldRejectUnderEnforcement(o)));
     }
 }

src/Backend/AutopilotMonitor.Functions.Tests/SharedManifestParityTests.cs

@@ -88,6 +88,8 @@ public sealed class SharedManifestParityTests
             ["webhookProviderTypes"] = EnumMap<WebhookProviderType>(),
             ["annotationLanes"] = AnnotationLanes.All,
             ["annotationVerdicts"] = AnnotationVerdicts.All,
+            // Install channels of the app rows (AppInstallSummary.Source and the per-app DTOs).
+            ["appInstallSources"] = AppInstallSources.All,
             // Community rule submissions: kinds, effective statuses (incl. the derived "published"),
             // attribution modes and reviewer decisions — the MCP tool schemas enumerate these.
             ["ruleSubmissionKinds"] = RuleSubmissionKinds.All,

src/Backend/AutopilotMonitor.Functions.Tests/SsrfGuardTests.cs

@@ -1,8 +1,14 @@
 using System.Net;
+using System.Text;
+using System.Text.RegularExpressions;
 using AutopilotMonitor.Functions.Security;
 
 namespace AutopilotMonitor.Functions.Tests;
 
+/// <summary>
+/// Format checks and the address classifier. The connection gate itself is covered on real sockets
+/// in <see cref="WebhookConnectGateTests"/>.
+/// </summary>
 public class SsrfGuardTests
 {
     // ── ValidateWebhookUrlFormat: valid URLs ──
@@ -145,37 +151,124 @@ public class SsrfGuardTests
         Assert.False(SsrfGuard.IsBlockedAddress(IPAddress.Parse(ip)));
     }
 
-    // ── ValidateDestinationAsync: integration ──
+    // ── IsBlockedAddress: IPv6 beyond the classic ranges ──
 
-    [Fact]
-    public async Task ValidateDestination_Localhost_ThrowsSsrfException()
+    [Theory]
+    [InlineData("::")]                     // unspecified
+    [InlineData("::ffff:8.8.8.8")]         // IPv4-mapped, even of a public address
+    [InlineData("::8.8.8.8")]              // IPv4-compatible (deprecated)
+    [InlineData("64:ff9b::a00:1")]         // NAT64 of 10.0.0.1
+    [InlineData("2002:a00:1::1")]          // 6to4 of 10.0.0.1
+    [InlineData("2001::1")]                // Teredo
+    [InlineData("2001:db8::1")]            // documentation
+    [InlineData("fec0::1")]                // deprecated site-local
+    [InlineData("4000::1")]                // outside global unicast
+    [InlineData("e000::1")]                // outside global unicast
+    public void IsBlocked_Ipv6OutsidePublicUnicast_ReturnsTrue(string ip)
     {
-        var ex = await Assert.ThrowsAsync<SsrfException>(
-            () => SsrfGuard.ValidateDestinationAsync("https://localhost/webhook"));
-        Assert.Contains("private or reserved", ex.Message);
+        Assert.True(SsrfGuard.IsBlockedAddress(IPAddress.Parse(ip)));
     }
 
-    [Fact]
-    public async Task ValidateDestination_HttpScheme_ThrowsSsrfException()
+    [Theory]
+    [InlineData("fe80::1%5")]
+    [InlineData("2606:4700:4700::1111%5")]
+    public void IsBlocked_Ipv6WithZoneId_ReturnsTrue(string ip)
     {
-        var ex = await Assert.ThrowsAsync<SsrfException>(
-            () => SsrfGuard.ValidateDestinationAsync("http://example.com/webhook"));
-        Assert.Contains("HTTPS", ex.Message);
+        Assert.True(SsrfGuard.IsBlockedAddress(IPAddress.Parse(ip)));
     }
 
-    [Fact]
-    public async Task ValidateDestination_InvalidUrl_ThrowsSsrfException()
+    [Theory]
+    [InlineData("2606:4700:4700::1111")]   // Cloudflare
+    [InlineData("2a00:1450:4001::1")]      // Google
+    [InlineData("2001:200::1")]            // just above 2001::/23
+    [InlineData("2003::1")]                // just above 6to4 2002::/16
+    [InlineData("2001:db9::1")]            // just above documentation 2001:db8::/32
+    [InlineData("192.0.1.0")]              // just above 192.0.0.0/24
+    [InlineData("198.20.0.0")]             // just above benchmarking 198.18.0.0/15
+    [InlineData("223.255.255.255")]        // just below multicast
+    public void IsBlocked_PublicNeighbourOfRefusedBlock_ReturnsFalse(string ip)
     {
-        await Assert.ThrowsAsync<SsrfException>(
-            () => SsrfGuard.ValidateDestinationAsync("not-a-url"));
+        Assert.False(SsrfGuard.IsBlockedAddress(IPAddress.Parse(ip)));
     }
 
+    // ── IsBlockedAddress: every IANA special-purpose block (tests/fixtures/iana-special-purpose-addresses) ──
+
+    private static readonly string[] RegistryFiles = { "iana-ipv4-special-registry-1.csv", "iana-ipv6-special-registry-1.csv" };
+
+    public static IEnumerable<object[]> RegistryBlocks() =>
+        RegistryFiles.SelectMany(file => ReadRegistryBlocks(file).Select(block => new object[] { file, block }));
+
     [Fact]
-    public async Task ValidateDestination_UnresolvableHost_ThrowsSsrfException()
+    public void Registry_IsReadCompletely()
     {
-        var ex = await Assert.ThrowsAsync<SsrfException>(
-            () => SsrfGuard.ValidateDestinationAsync("https://this-host-does-not-exist-xyz123.invalid/webhook"));
-        Assert.Contains("resolve", ex.Message);
+        foreach (var file in RegistryFiles)
+        {
+            var blocks = ReadRegistryBlocks(file);
+            Assert.True(blocks.Count >= 20, $"{file}: only {blocks.Count} blocks read");
+            Assert.All(blocks, block => Assert.Matches(@"^[0-9a-f.:]+/\d{1,3}$", block));
+        }
+    }
+
+    [Theory]
+    [MemberData(nameof(RegistryBlocks))]
+    public void IsBlocked_EveryIanaSpecialPurposeBlock_FirstAndLastAddress(string file, string block)
+    {
+        var network = IPNetwork.Parse(block);
+        var first = network.BaseAddress.GetAddressBytes();
+        var last = (byte[])first.Clone();
+        for (var bit = network.PrefixLength; bit < last.Length * 8; bit++)
+            last[bit / 8] |= (byte)(0x80 >> (bit % 8));
+
+        foreach (var ip in new[] { new IPAddress(first), new IPAddress(last) })
+            Assert.True(SsrfGuard.IsBlockedAddress(ip), $"{file}: {block} -> {ip}");
+    }
+
+    private static List<string> ReadRegistryBlocks(string file)
+    {
+        var dir = new DirectoryInfo(AppContext.BaseDirectory);
+        while (dir != null && !File.Exists(Path.Combine(dir.FullName, "AutopilotMonitor.sln")))
+            dir = dir.Parent;
+        Assert.NotNull(dir);
+        var rows = ParseCsv(File.ReadAllText(Path.Combine(dir!.FullName, "tests", "fixtures", "iana-special-purpose-addresses", file)));
+        Assert.Equal("Address Block", rows[0][0]);
+        // Footnote markers ("2002::/16 [3]") dropped, cells that list two blocks split.
+        return rows.Skip(1)
+            .SelectMany(row => row[0].Split(','))
+            .Select(block => Regex.Replace(block, @"\[\d+\]", "").Trim())
+            .Where(block => block.Length > 0)
+            .ToList();
+    }
+
+    /// <summary>RFC 4180 reader — the registry quotes cells, wraps the RFC column over lines and doubles quotes.</summary>
+    private static List<List<string>> ParseCsv(string text)
+    {
+        var rows = new List<List<string>>();
+        var row = new List<string>();
+        var cell = new StringBuilder();
+        var quoted = false;
+        for (var i = 0; i < text.Length; i++)
+        {
+            var c = text[i];
+            if (quoted)
+            {
+                if (c == '"' && i + 1 < text.Length && text[i + 1] == '"') { cell.Append('"'); i++; }
+                else if (c == '"') quoted = false;
+                else cell.Append(c);
+            }
+            else if (c == '"') quoted = true;
+            else if (c == ',') { row.Add(cell.ToString()); cell.Clear(); }
+            else if (c == '\n' || c == '\r')
+            {
+                if (c == '\r' && i + 1 < text.Length && text[i + 1] == '\n') i++;
+                row.Add(cell.ToString());
+                rows.Add(row);
+                row = new List<string>();
+                cell.Clear();
+            }
+            else cell.Append(c);
+        }
+        if (cell.Length > 0 || row.Count > 0) { row.Add(cell.ToString()); rows.Add(row); }
+        return rows;
     }
 
     // ── ValidateAzureBlobSasUrlFormat ──

src/Backend/AutopilotMonitor.Functions.Tests/TenantConfigUpdateTests.cs

@@ -0,0 +1,368 @@
+using System.Net;
+using AutopilotMonitor.Functions.Functions.Admin;
+using AutopilotMonitor.Functions.Functions.Config;
+using AutopilotMonitor.Functions.Services;
+using AutopilotMonitor.Shared;
+using AutopilotMonitor.Shared.DataAccess;
+using AutopilotMonitor.Shared.Models;
+using Microsoft.Extensions.Caching.Memory;
+using Microsoft.Extensions.Logging.Abstractions;
+using Moq;
+
+namespace AutopilotMonitor.Functions.Tests;
+
+/// <summary>
+/// <see cref="TenantConfigurationService.UpdateAsync"/> and the plan writers built on it: the decision runs
+/// on the fresh row and the write is conditional on that row's ETag — a cached view (up to 5 minutes stale
+/// per instance) is never saved, so it can neither rewind another writer's change nor lift the offboarding
+/// tombstone during the drain barrier, and two concurrent trial starts cannot both see TrialConsumed=false.
+/// </summary>
+public class TenantConfigUpdateTests
+{
+    private const string TenantId = "aaaaaaaa-bbbb-cccc-dddd-eeeeeeeeeeee";
+
+    private readonly Mock<IConfigRepository> _repo = new();
+    private readonly Queue<(TenantConfiguration Row, string ETag)> _freshReads = new();
+    private readonly List<(TenantConfiguration Config, string ETag, string? Source, string? Reason)> _replaces = new();
+    private readonly Queue<bool> _replaceResults = new();
+    private readonly List<TenantConfiguration> _creates = new();
+    private readonly Queue<bool> _createResults = new();
+
+    public TenantConfigUpdateTests()
+    {
+        _repo.Setup(r => r.GetTenantConfigurationWithEtagAsync(TenantId))
+            .ReturnsAsync(() => _freshReads.Count > 0 ? _freshReads.Dequeue() : null);
+        _repo.Setup(r => r.TryReplaceTenantConfigurationAsync(
+                It.IsAny<TenantConfiguration>(), It.IsAny<string>(), It.IsAny<string?>(), It.IsAny<string?>()))
+            .Callback<TenantConfiguration, string, string?, string?>((c, e, s, r) => _replaces.Add((c, e, s, r)))
+            .ReturnsAsync(() => _replaceResults.Count == 0 || _replaceResults.Dequeue());
+        _repo.Setup(r => r.TryCreateTenantConfigurationAsync(It.IsAny<TenantConfiguration>()))
+            .Callback<TenantConfiguration>(c => _creates.Add(c))
+            .ReturnsAsync(() => _createResults.Count == 0 || _createResults.Dequeue());
+    }
+
+    private TenantConfigurationService Service(ManagedTenantProIndex? proIndex = null) =>
+        new(_repo.Object, NullLogger<TenantConfigurationService>.Instance,
+            new MemoryCache(new MemoryCacheOptions()), proIndex ?? ManagedTenantProIndex.None);
+
+    private static TenantConfiguration Row(Action<TenantConfiguration>? shape = null)
+    {
+        var row = TenantConfiguration.CreateDefault(TenantId);
+        row.DomainName = "contoso.com";
+        row.ContactEmail = "it@contoso.com";
+        row.CompanyName = "Contoso";
+        shape?.Invoke(row);
+        return row;
+    }
+
+    private static TenantConfiguration Tombstone() => Row(r =>
+    {
+        r.Disabled = true;
+        r.DisabledReason = TenantOffboardFunction.OffboardingDisabledReason;
+    });
+
+    /// <summary>Primes the service's cache with <paramref name="cached"/>, the view a warm instance would hold.</summary>
+    private async Task PrimeCacheAsync(TenantConfigurationService service, TenantConfiguration cached)
+    {
+        _repo.Setup(r => r.GetTenantConfigurationAsync(TenantId)).ReturnsAsync(cached);
+        Assert.Same(cached, await service.GetConfigurationIfExistsAsync(TenantId));
+    }
+
+    // ── UpdateAsync ─────────────────────────────────────────────────────────
+
+    [Fact]
+    public async Task UpdateAsync_MutatesTheFreshRow_UnderItsETag_NeverTheCachedView()
+    {
+        var service = Service();
+        await PrimeCacheAsync(service, Row(r => r.HomedAppClientId = null));
+        _freshReads.Enqueue((Row(r => r.HomedAppClientId = "flipped-meanwhile"), "etag-1"));
+
+        var update = await service.UpdateAsync(TenantId, row => { row.PayingCustomer = true; return true; }, "plan", "why");
+
+        Assert.Equal(TenantConfigUpdateStatus.Updated, update.Status);
+        var (written, etag, source, reason) = Assert.Single(_replaces);
+        Assert.Equal("etag-1", etag);
+        Assert.Equal("flipped-meanwhile", written.HomedAppClientId);
+        Assert.True(written.PayingCustomer);
+        Assert.Equal("plan", source);
+        Assert.Equal("why", reason);
+
+        // The cached view is gone: the next read goes to storage.
+        await service.GetConfigurationIfExistsAsync(TenantId);
+        _repo.Verify(r => r.GetTenantConfigurationAsync(TenantId), Times.Exactly(2));
+    }
+
+    [Fact]
+    public async Task UpdateAsync_LostRace_RereadsAndDecidesAgainOnTheNewRow()
+    {
+        var first = Row();
+        var second = Row(r => r.CompanyName = "Contoso after the concurrent write");
+        _freshReads.Enqueue((first, "etag-1"));
+        _freshReads.Enqueue((second, "etag-2"));
+        _replaceResults.Enqueue(false);
+        _replaceResults.Enqueue(true);
+        var seen = new List<TenantConfiguration>();
+
+        var update = await Service().UpdateAsync(TenantId, row => { seen.Add(row); return true; }, "plan", "why");
+
+        Assert.Equal(TenantConfigUpdateStatus.Updated, update.Status);
+        Assert.Equal(new[] { first, second }, seen);
+        Assert.Equal(new[] { "etag-1", "etag-2" }, _replaces.Select(r => r.ETag));
+        Assert.Same(second, update.Config);
+    }
+
+    [Fact]
+    public async Task UpdateAsync_EveryAttemptLosesTheRace_Conflict()
+    {
+        for (var i = 0; i < TenantConfigurationService.MaxUpdateAttempts; i++)
+        {
+            _freshReads.Enqueue((Row(), $"etag-{i}"));
+            _replaceResults.Enqueue(false);
+        }
+
+        var update = await Service().UpdateAsync(TenantId, _ => true, "plan", "why");
+
+        Assert.Equal(TenantConfigUpdateStatus.Conflict, update.Status);
+        Assert.Equal(TenantConfigurationService.MaxUpdateAttempts, _replaces.Count);
+    }
+
+    [Fact]
+    public async Task UpdateAsync_OffboardingTombstone_IsNeverWritten()
+    {
+        _freshReads.Enqueue((Tombstone(), "etag-1"));
+        var mutated = false;
+
+        var update = await Service().UpdateAsync(TenantId, _ => mutated = true, "plan", "why");
+
+        Assert.Equal(TenantConfigUpdateStatus.OffboardingInProgress, update.Status);
+        Assert.False(mutated);
+        Assert.Empty(_replaces);
+    }
+
+    [Fact]
+    public async Task UpdateAsync_MutationDeclines_NothingIsWritten()
+    {
+        var row = Row();
+        _freshReads.Enqueue((row, "etag-1"));
+
+        var update = await Service().UpdateAsync(TenantId, _ => false, "plan", "why");
+
+        Assert.Equal(TenantConfigUpdateStatus.Declined, update.Status);
+        Assert.Same(row, update.Config);
+        Assert.Empty(_replaces);
+    }
+
+    [Fact]
+    public async Task UpdateAsync_NoRow_NotFound()
+    {
+        var update = await Service().UpdateAsync(TenantId, _ => true, "plan", "why");
+
+        Assert.Equal(TenantConfigUpdateStatus.NotFound, update.Status);
+        Assert.Empty(_replaces);
+    }
+
+    [Fact]
+    public async Task UpdateAsync_TheMutationSeesTheConferredProProjection()
+    {
+        // EvaluateTrialStart's AlreadyPro verdict for a managed tenant depends on it.
+        _freshReads.Enqueue((Row(), "etag-1"));
+        string? seen = null;
+
+        await Service(new StubManagedTenantProIndex(_ => "managing-tenant"))
+            .UpdateAsync(TenantId, row => { seen = row.ManagedByProTenantId; return false; }, "plan", "why");
+
+        Assert.Equal("managing-tenant", seen);
+    }
+
+    // ── CreateOrUpdateAsync + the default row ───────────────────────────────
+
+    [Fact]
+    public async Task CreateOrUpdateAsync_NoRow_InsertsTheMutatedDefault()
+    {
+        var update = await Service().CreateOrUpdateAsync(TenantId, row => { row.DomainName = "contoso.com"; return true; }, "auth", "why");
+
+        Assert.Equal(TenantConfigUpdateStatus.Updated, update.Status);
+        var created = Assert.Single(_creates);
+        Assert.Equal("contoso.com", created.DomainName);
+        Assert.Equal(TenantId, created.TenantId);
+        Assert.Empty(_replaces);
+    }
+
+    [Fact]
+    public async Task CreateOrUpdateAsync_RowCreatedConcurrently_DecidesAgainOnTheStoredRow()
+    {
+        // Our read saw no row, the insert lost (409), the re-read finds the concurrent writer's row.
+        var stored = Row(r => r.DomainName = "seeded-by-the-other-login.com");
+        var reads = 0;
+        _repo.Setup(r => r.GetTenantConfigurationWithEtagAsync(TenantId))
+            .ReturnsAsync(() => reads++ == 0 ? null : (stored, "etag-2"));
+        _createResults.Enqueue(false);
+        var seen = new List<string>();
+
+        var update = await Service().CreateOrUpdateAsync(TenantId, row =>
+        {
+            seen.Add(row.DomainName);
+            return string.IsNullOrEmpty(row.DomainName);
+        }, "auth", "why");
+
+        Assert.Equal(TenantConfigUpdateStatus.Declined, update.Status);
+        Assert.Equal(new[] { "", "seeded-by-the-other-login.com" }, seen);
+        Assert.Single(_creates);
+        Assert.Empty(_replaces);
+    }
+
+    [Fact]
+    public async Task CreateOrUpdateAsync_Tombstone_IsNeverWritten()
+    {
+        _freshReads.Enqueue((Tombstone(), "etag-1"));
+
+        var update = await Service().CreateOrUpdateAsync(TenantId, _ => true, "offboard", "why");
+
+        Assert.Equal(TenantConfigUpdateStatus.OffboardingInProgress, update.Status);
+        _repo.VerifyNoTenantConfigWrite();
+    }
+
+    [Fact]
+    public async Task GetConfigurationAsync_NoRow_CreatesTheDefaultConditionally()
+    {
+        _repo.Setup(r => r.GetTenantConfigurationAsync(TenantId)).ReturnsAsync((TenantConfiguration?)null);
+
+        var config = await Service().GetConfigurationAsync(TenantId);
+
+        Assert.Same(Assert.Single(_creates), config);
+        _repo.Verify(r => r.TryReplaceTenantConfigurationAsync(
+            It.IsAny<TenantConfiguration>(), It.IsAny<string>(), It.IsAny<string?>(), It.IsAny<string?>()), Times.Never);
+    }
+
+    [Fact]
+    public async Task GetConfigurationAsync_DefaultLostTheRace_ReturnsTheRowTheOtherWriterCreated()
+    {
+        // The first-login seed landed between our read and our insert: never overwrite it.
+        var seeded = Row(r => r.OnboardedBy = "first@contoso.com");
+        var reads = 0;
+        _repo.Setup(r => r.GetTenantConfigurationAsync(TenantId))
+            .ReturnsAsync(() => reads++ == 0 ? null : seeded);
+        _createResults.Enqueue(false);
+
+        var config = await Service().GetConfigurationAsync(TenantId);
+
+        Assert.Same(seeded, config);
+        Assert.Single(_creates);
+        _repo.Verify(r => r.TryReplaceTenantConfigurationAsync(
+            It.IsAny<TenantConfiguration>(), It.IsAny<string>(), It.IsAny<string?>(), It.IsAny<string?>()), Times.Never);
+    }
+
+    // ── Endpoints ───────────────────────────────────────────────────────────
+
+    private (PlanManagementFunction Sut, Mock<IOpsEventRepository> OpsRepo, Mock<IMaintenanceRepository> Audit) Plan(
+        TenantConfigurationService service)
+    {
+        var adminConfig = new Mock<AdminConfigurationService>(
+            _repo.Object, NullLogger<AdminConfigurationService>.Instance, new MemoryCache(new MemoryCacheOptions()))
+        { CallBase = false };
+        var opsRepo = new Mock<IOpsEventRepository>();
+        var audit = new Mock<IMaintenanceRepository>();
+        var sut = new PlanManagementFunction(
+            NullLogger<PlanManagementFunction>.Instance, service, adminConfig.Object, audit.Object,
+            new OpsEventService(opsRepo.Object, NullLogger<OpsEventService>.Instance,
+                TestNotifications.InertOpsAlertDispatch(adminConfig.Object)),
+            TestProConferral.Inert());
+        return (sut, opsRepo, audit);
+    }
+
+    [Fact]
+    public async Task StartTrial_ConcurrentStartWonTheRace_SecondGets409TrialAlreadyConsumed()
+    {
+        _freshReads.Enqueue((Row(), "etag-1"));
+        _freshReads.Enqueue((Row(r => r.TrialConsumed = true), "etag-2"));
+        _replaceResults.Enqueue(false);
+        var (sut, opsRepo, audit) = Plan(Service());
+        var (req, _) = EndpointHarness.Request(TenantId);
+
+        var response = await sut.StartTrial(req, TenantId);
+
+        Assert.Equal(HttpStatusCode.Conflict, response.StatusCode);
+        Assert.Equal(Constants.ApiErrorCodes.TrialAlreadyConsumed, EndpointHarness.ErrorCode(response));
+        Assert.Single(_replaces);
+        opsRepo.Verify(r => r.SaveOpsEventAsync(It.IsAny<OpsEventEntry>()), Times.Never);
+        audit.Verify(a => a.LogAuditEntryAsync(It.IsAny<string>(), It.IsAny<string>(), It.IsAny<string>(),
+            It.IsAny<string>(), It.IsAny<string>(), It.IsAny<Dictionary<string, string>>()), Times.Never);
+    }
+
+    [Fact]
+    public async Task StartTrial_WarmCacheDuringTheDrainBarrier_CannotLiftTheTombstone()
+    {
+        var service = Service();
+        await PrimeCacheAsync(service, Row());      // this instance still sees the tenant enabled
+        _freshReads.Enqueue((Tombstone(), "etag-1"));
+        var (sut, _, _) = Plan(service);
+        var (req, _) = EndpointHarness.Request(TenantId);
+
+        var response = await sut.StartTrial(req, TenantId);
+
+        Assert.Equal(HttpStatusCode.Forbidden, response.StatusCode);
+        Assert.Equal(Constants.ApiErrorCodes.TenantSuspended, EndpointHarness.ErrorCode(response));
+        Assert.Empty(_replaces);
+        _repo.VerifyNoTenantConfigWrite();
+    }
+
+    [Fact]
+    public async Task StartTrial_Allowed_WritesTheTrialOntoTheFreshRow()
+    {
+        var service = Service();
+        await PrimeCacheAsync(service, Row(r => r.CompanyName = "stale"));
+        _freshReads.Enqueue((Row(r => r.CompanyName = "Contoso"), "etag-1"));
+        var (sut, opsRepo, _) = Plan(service);
+        var (req, _) = EndpointHarness.Request(TenantId, upn: "admin@contoso.com");
+
+        var response = await sut.StartTrial(req, TenantId);
+
+        Assert.Equal(HttpStatusCode.OK, response.StatusCode);
+        var (written, etag, source, reason) = Assert.Single(_replaces);
+        Assert.Equal("etag-1", etag);
+        Assert.Equal("Contoso", written.CompanyName);
+        Assert.True(written.TrialConsumed);
+        Assert.Equal("admin@contoso.com", written.TrialGrantedBy);
+        Assert.Equal(written.TrialStartedUtc!.Value.AddDays(PlanManagementFunction.SelfServiceTrialDays), written.TrialExpiresUtc);
+        Assert.Equal("plan", source);
+        Assert.Equal("self-service trial start", reason);
+        opsRepo.Verify(r => r.SaveOpsEventAsync(It.Is<OpsEventEntry>(e => e.EventType == OpsEventTypes.TenantTrialStarted)), Times.Once);
+    }
+
+    [Fact]
+    public async Task SetPlanTier_WritesOnTheFreshRowUnderItsETag()
+    {
+        var service = Service();
+        await PrimeCacheAsync(service, Row(r => r.CompanyName = "stale"));
+        _freshReads.Enqueue((Row(r => r.CompanyName = "Contoso"), "etag-1"));
+        var (sut, _, audit) = Plan(service);
+        var (req, _) = EndpointHarness.Request(TenantId,
+            jsonBody: $"{{\"{PlanManagementFunction.PlanPatchKeys.PayingCustomer}\":true}}");
+
+        var response = await sut.SetPlanTier(req, TenantId);
+
+        Assert.Equal(HttpStatusCode.OK, response.StatusCode);
+        var (written, etag, _, _) = Assert.Single(_replaces);
+        Assert.Equal("etag-1", etag);
+        Assert.Equal("Contoso", written.CompanyName);
+        Assert.True(written.PayingCustomer);
+        audit.Verify(a => a.LogAuditEntryAsync(TenantId, "UPDATE", "TenantPlan", TenantId, It.IsAny<string>(),
+            It.Is<Dictionary<string, string>>(d => d.ContainsKey("PayingCustomer"))), Times.Once);
+    }
+
+    [Fact]
+    public async Task SetPlanTier_NothingChangesOnTheFreshRow_200WithoutAWrite()
+    {
+        _freshReads.Enqueue((Row(r => r.PayingCustomer = true), "etag-1"));
+        var (sut, _, audit) = Plan(Service());
+        var (req, _) = EndpointHarness.Request(TenantId,
+            jsonBody: $"{{\"{PlanManagementFunction.PlanPatchKeys.PayingCustomer}\":true}}");
+
+        var response = await sut.SetPlanTier(req, TenantId);
+
+        Assert.Equal(HttpStatusCode.OK, response.StatusCode);
+        Assert.Empty(_replaces);
+        audit.Verify(a => a.LogAuditEntryAsync(It.IsAny<string>(), It.IsAny<string>(), It.IsAny<string>(),
+            It.IsAny<string>(), It.IsAny<string>(), It.IsAny<Dictionary<string, string>>()), Times.Never);
+    }
+}

src/Backend/AutopilotMonitor.Functions.Tests/TenantConfigurationServiceReadSemanticsTests.cs

@@ -100,7 +100,7 @@ public sealed class TenantConfigurationServiceReadSemanticsTests
         // row can never be Pro, whatever the index says (fail-closed).
         var (service, repo, _) = BuildProjecting(Owner);
         repo.Setup(r => r.GetTenantConfigurationAsync(TenantId)).ReturnsAsync((TenantConfiguration?)null);
-        repo.Setup(r => r.SaveTenantConfigurationAsync(It.IsAny<TenantConfiguration>())).ReturnsAsync(true);
+        repo.Setup(r => r.TryCreateTenantConfigurationAsync(It.IsAny<TenantConfiguration>())).ReturnsAsync(true);
 
         var config = await service.GetConfigurationAsync(TenantId);
 
@@ -135,7 +135,7 @@ public sealed class TenantConfigurationServiceReadSemanticsTests
             .ReturnsAsync((TenantConfiguration?)null);
 
         Assert.Null(await service.GetConfigurationIfExistsAsync(TenantId));
-        // MockBehavior.Strict: any SaveTenantConfigurationAsync call would have thrown.
+        // MockBehavior.Strict: any write (create or replace) would have thrown.
     }
 
     [Fact]
@@ -186,56 +186,60 @@ public sealed class TenantConfigurationServiceReadSemanticsTests
         Assert.Equal(TenantId, config.TenantId);
     }
 
-    // ── Save semantics (Codex finding 2026-07-07): the repository swallows storage exceptions
-    //    and reports failure via its bool return — the service must THROW on false so callers
-    //    (plan/trial endpoints, config PUT) can never audit + 200 a write that never persisted.
+    // ── Write semantics: UpdateAsync is the write path (D-290). A storage failure propagates, and
+    //    the cache is dropped on every attempt — success or failure — so no instance that was read
+    //    before the write is served afterwards as if it were current.
 
-    [Fact]
-    public async Task SaveConfigurationAsync_RepoReportsFalse_Throws()
+    private static Mock<IConfigRepository> WithStoredRow(Mock<IConfigRepository> repo)
     {
-        var (service, repo) = Build();
-        repo.Setup(r => r.SaveTenantConfigurationAsync(It.IsAny<TenantConfiguration>(), It.IsAny<string?>(), It.IsAny<string?>()))
-            .ReturnsAsync(false);
-
-        await Assert.ThrowsAsync<InvalidOperationException>(
-            () => service.SaveConfigurationAsync(new TenantConfiguration { TenantId = TenantId, UpdatedBy = "alice@contoso.com" }));
+        repo.Setup(r => r.GetTenantConfigurationAsync(TenantId))
+            .ReturnsAsync(new TenantConfiguration { TenantId = TenantId, PlanTier = "community", UpdatedBy = "x" });
+        repo.Setup(r => r.GetTenantConfigurationWithEtagAsync(TenantId))
+            .ReturnsAsync((new TenantConfiguration { TenantId = TenantId, PlanTier = "community", UpdatedBy = "x" }, "etag"));
+        return repo;
     }
 
     [Fact]
-    public async Task SaveConfigurationAsync_FailedSave_InvalidatesCachedInstance()
+    public async Task UpdateAsync_StorageError_Propagates()
     {
-        // The plan/trial endpoints mutate the CACHED instance in place before saving. A failed
-        // save must drop that instance from the cache — otherwise the unsaved mutation is served
-        // as if persisted for up to 5 minutes.
         var (service, repo) = Build();
-        var stored = new TenantConfiguration { TenantId = TenantId, PlanTier = "free", UpdatedBy = "x" };
-        repo.Setup(r => r.GetTenantConfigurationAsync(TenantId)).ReturnsAsync(stored);
-        repo.Setup(r => r.SaveTenantConfigurationAsync(It.IsAny<TenantConfiguration>(), It.IsAny<string?>(), It.IsAny<string?>()))
-            .ReturnsAsync(false);
+        WithStoredRow(repo).Setup(r => r.TryReplaceTenantConfigurationAsync(
+                It.IsAny<TenantConfiguration>(), "etag", It.IsAny<string?>(), It.IsAny<string?>()))
+            .ThrowsAsync(new InvalidOperationException("throttled"));
 
-        var cached = await service.GetConfigurationIfExistsAsync(TenantId);
-        cached!.PlanTier = "enterprise"; // in-place mutation, as the plan endpoint does
+        await Assert.ThrowsAsync<InvalidOperationException>(
+            () => service.UpdateAsync(TenantId, row => { row.PlanTier = "pro"; return true; }, "plan", "test"));
+    }
 
-        await Assert.ThrowsAsync<InvalidOperationException>(() => service.SaveConfigurationAsync(cached));
+    [Fact]
+    public async Task UpdateAsync_FailedWrite_InvalidatesTheCache()
+    {
+        var (service, repo) = Build();
+        WithStoredRow(repo).Setup(r => r.TryReplaceTenantConfigurationAsync(
+                It.IsAny<TenantConfiguration>(), "etag", It.IsAny<string?>(), It.IsAny<string?>()))
+            .ThrowsAsync(new InvalidOperationException("throttled"));
+
+        await service.GetConfigurationIfExistsAsync(TenantId); // prime cache
+        await Assert.ThrowsAsync<InvalidOperationException>(
+            () => service.UpdateAsync(TenantId, row => { row.PlanTier = "pro"; return true; }, "plan", "test"));
+        await service.GetConfigurationIfExistsAsync(TenantId); // must go back to storage
 
-        // Next read must go back to storage (cache dropped), not serve the mutated instance.
-        await service.GetConfigurationIfExistsAsync(TenantId);
         repo.Verify(r => r.GetTenantConfigurationAsync(TenantId), Times.Exactly(2));
     }
 
     [Fact]
-    public async Task SaveConfigurationAsync_Success_InvalidatesCache()
+    public async Task UpdateAsync_Success_InvalidatesTheCache_AndNeverTouchesTheCachedInstance()
     {
         var (service, repo) = Build();
-        repo.Setup(r => r.GetTenantConfigurationAsync(TenantId))
-            .ReturnsAsync(new TenantConfiguration { TenantId = TenantId, UpdatedBy = "x" });
-        repo.Setup(r => r.SaveTenantConfigurationAsync(It.IsAny<TenantConfiguration>(), It.IsAny<string?>(), It.IsAny<string?>()))
+        WithStoredRow(repo).Setup(r => r.TryReplaceTenantConfigurationAsync(
+                It.IsAny<TenantConfiguration>(), "etag", It.IsAny<string?>(), It.IsAny<string?>()))
             .ReturnsAsync(true);
 
-        await service.GetConfigurationIfExistsAsync(TenantId); // prime cache
-        await service.SaveConfigurationAsync(new TenantConfiguration { TenantId = TenantId, UpdatedBy = "x" });
-        await service.GetConfigurationIfExistsAsync(TenantId); // must re-read after save
+        var cached = await service.GetConfigurationIfExistsAsync(TenantId); // prime cache
+        await service.UpdateAsync(TenantId, row => { row.PlanTier = "pro"; return true; }, "plan", "test");
+        await service.GetConfigurationIfExistsAsync(TenantId); // must re-read after the write
 
+        Assert.Equal("community", cached!.PlanTier);
         repo.Verify(r => r.GetTenantConfigurationAsync(TenantId), Times.Exactly(2));
     }
 }

src/Backend/AutopilotMonitor.Functions.Tests/TenantContactEmailTests.cs

@@ -43,7 +43,7 @@ public class TenantContactEmailTests
     }
 
     // ------------------------------------------------------------------
-    // Server-side validation (UpdateTenantConfigurationFunction)
+    // Server-side validation (TenantConfigValidation)
     // ------------------------------------------------------------------
 
     [Theory]
@@ -51,7 +51,7 @@ public class TenantContactEmailTests
     [InlineData("first.last+autopilot@sub.fabrikam.co.uk")]
     [InlineData("UPPER@CONTOSO.COM")]
     public void ValidateContactEmail_accepts_real_addresses(string email)
-        => Assert.Null(UpdateTenantConfigurationFunction.ValidateContactEmail(email));
+        => Assert.Null(TenantConfigValidation.ValidateContactEmail(email));
 
     [Theory]
     [InlineData(null)]
@@ -60,7 +60,7 @@ public class TenantContactEmailTests
     public void ValidateContactEmail_treats_absence_as_valid(string? email)
     {
         // No contact address is a legitimate state — it means we cannot reach this tenant.
-        Assert.Null(UpdateTenantConfigurationFunction.ValidateContactEmail(email));
+        Assert.Null(TenantConfigValidation.ValidateContactEmail(email));
     }
 
     [Theory]
@@ -76,7 +76,7 @@ public class TenantContactEmailTests
     [InlineData("Ops Team <ops@contoso.com>", "single address")]          // display-name form
     public void ValidateContactEmail_rejects_values_that_are_not_a_single_address(string email, string expectedHint)
     {
-        var error = UpdateTenantConfigurationFunction.ValidateContactEmail(email);
+        var error = TenantConfigValidation.ValidateContactEmail(email);
         Assert.NotNull(error);
         Assert.Contains(expectedHint, error!, StringComparison.OrdinalIgnoreCase);
     }
@@ -85,7 +85,7 @@ public class TenantContactEmailTests
     public void ValidateContactEmail_rejects_header_injection()
     {
         // Once this address is actually mailed, an embedded CR/LF would forge mail headers.
-        var error = UpdateTenantConfigurationFunction.ValidateContactEmail(
+        var error = TenantConfigValidation.ValidateContactEmail(
             $"ops@contoso.com{(char)13}{(char)10}Bcc: attacker@evil.test");
 
         Assert.NotNull(error);
@@ -94,8 +94,8 @@ public class TenantContactEmailTests
     [Fact]
     public void ValidateContactEmail_rejects_an_over_long_value()
     {
-        var local = new string('a', UpdateTenantConfigurationFunction.MaxContactEmailLength);
-        var error = UpdateTenantConfigurationFunction.ValidateContactEmail($"{local}@contoso.com");
+        var local = new string('a', TenantConfigValidation.MaxContactEmailLength);
+        var error = TenantConfigValidation.ValidateContactEmail($"{local}@contoso.com");
 
         Assert.NotNull(error);
         Assert.Contains("at most", error!);
@@ -189,9 +189,7 @@ public class TenantContactEmailTests
 
         await service.TrySeedContactEmailAsync(TenantId, "ops@contoso.com");
 
-        repo.Verify(r => r.SaveTenantConfigurationAsync(It.IsAny<TenantConfiguration>()), Times.Never);
-        repo.Verify(r => r.SaveTenantConfigurationAsync(
-            It.IsAny<TenantConfiguration>(), It.IsAny<string?>(), It.IsAny<string?>()), Times.Never);
+        repo.VerifyNoTenantConfigWrite();
     }
 
     // ------------------------------------------------------------------

src/Backend/AutopilotMonitor.Functions.Tests/TenantEntitlementServiceTests.cs

@@ -86,9 +86,7 @@ public class TenantEntitlementServiceTests
         // auto-create + persist a default config row.
         var (svc, repo) = Build(config: null);
         await svc.GetEditionAsync(TenantId);
-        repo.Verify(r => r.SaveTenantConfigurationAsync(It.IsAny<TenantConfiguration>()), Times.Never);
-        repo.Verify(r => r.SaveTenantConfigurationAsync(
-            It.IsAny<TenantConfiguration>(), It.IsAny<string?>(), It.IsAny<string?>()), Times.Never);
+        repo.VerifyNoTenantConfigWrite();
     }
 
     [Fact]

src/Backend/AutopilotMonitor.Functions.Tests/TenantOffboardFunctionTests.cs

@@ -52,9 +52,7 @@ public sealed class TenantOffboardFunctionTests
         alreadyDisabled.Disabled = true;
         alreadyDisabled.DisabledReason = "Offboarding in progress";
         alreadyDisabled.LastUpdated = DateTime.UtcNow.AddDays(-30);
-        configRepoMock
-            .Setup(r => r.GetTenantConfigurationAsync(It.IsAny<string>()))
-            .ReturnsAsync(alreadyDisabled);
+        configRepoMock.BackWithRow(alreadyDisabled);
 
         var cache = new Microsoft.Extensions.Caching.Memory.MemoryCache(
             new Microsoft.Extensions.Caching.Memory.MemoryCacheOptions());
@@ -196,7 +194,7 @@ public sealed class TenantOffboardFunctionTests
         var sut = Build(repo);
         var historyRowKey = $"{DateTime.UtcNow:yyyyMMddHHmmssfff}_{TenantId}";
 
-        await sut.UpsertPointerWithCasAsync(TenantId, historyRowKey, DateTime.UtcNow);
+        await sut.UpsertPointerWithCasAsync(TenantId, historyRowKey, DateTime.UtcNow, trialConsumed: false);
 
         Assert.True(repo.Pointers.TryGetValue(TenantId, out var entry));
         Assert.Equal(historyRowKey, entry.Pointer.LatestHistoryRowKey);
@@ -222,13 +220,108 @@ public sealed class TenantOffboardFunctionTests
         var sut = Build(repo);
         var newHistoryRowKey = $"{DateTime.UtcNow:yyyyMMddHHmmssfff}_{TenantId}";
 
-        await sut.UpsertPointerWithCasAsync(TenantId, newHistoryRowKey, DateTime.UtcNow);
+        await sut.UpsertPointerWithCasAsync(TenantId, newHistoryRowKey, DateTime.UtcNow, trialConsumed: false);
 
         Assert.Equal(newHistoryRowKey, repo.Pointers[TenantId].Pointer.LatestHistoryRowKey);
         Assert.Equal("Initiated", repo.Pointers[TenantId].Pointer.LatestStatus);
         Assert.Equal(3, repo.Pointers[TenantId].Pointer.OffboardCount);
     }
 
+    [Theory]
+    [InlineData(false, false, false)]
+    [InlineData(false, true, true)]
+    [InlineData(true, false, true)]   // a later round without a trial never clears it
+    [InlineData(true, true, true)]
+    public async Task UpsertPointer_TrialConsumed_IsStickyAcrossRounds(bool stored, bool thisRound, bool expected)
+    {
+        var repo = new FakeOffboardingAuditRepository();
+        await repo.InsertByTenantPointerAsync(new OffboardingByTenantPointer
+        {
+            PartitionKey = Constants.OffboardingPartitionKeys.ByTenant,
+            RowKey = TenantId,
+            TenantId = TenantId,
+            LatestHistoryRowKey = "20260101000000000_" + TenantId,
+            LatestStatus = "Completed",
+            OffboardCount = 1,
+            TrialConsumed = stored,
+        });
+
+        await Build(repo).UpsertPointerWithCasAsync(TenantId, "20260201000000000_" + TenantId, DateTime.UtcNow, thisRound);
+
+        Assert.Equal(expected, repo.Pointers[TenantId].Pointer.TrialConsumed);
+        Assert.Equal(2, repo.Pointers[TenantId].Pointer.OffboardCount);
+    }
+
+    [Theory]
+    [InlineData(true)]
+    [InlineData(false)]
+    public async Task UpsertPointer_FreshTenant_StampsTrialConsumed(bool trialConsumed)
+    {
+        var repo = new FakeOffboardingAuditRepository();
+
+        await Build(repo).UpsertPointerWithCasAsync(TenantId, "20260201000000000_" + TenantId, DateTime.UtcNow, trialConsumed);
+
+        Assert.Equal(trialConsumed, repo.Pointers[TenantId].Pointer.TrialConsumed);
+    }
+
+    // ── OffboardTenant (Phase 1): the consumed trial must reach the pointer before anything is deleted ──
+
+    private static (TenantOffboardFunction Sut, FakeOffboardingAuditRepository Repo, RecordingEnqueuer Enqueuer, Mock<IConfigRepository> ConfigRepo)
+        BuildForOffboard(TenantConfiguration? liveConfig, Exception? readFailure = null)
+    {
+        var configRepo = new Mock<IConfigRepository>();
+        configRepo.BackWithRow(liveConfig).ThrowOnRead = readFailure;
+
+        var repo = new FakeOffboardingAuditRepository();
+        var enqueuer = new RecordingEnqueuer();
+        var sut = new TenantOffboardFunction(
+            logger: NullLogger<TenantOffboardFunction>.Instance,
+            configRepo: configRepo.Object,
+            tenantConfigService: new TenantConfigurationService(
+                configRepo.Object, NullLogger<TenantConfigurationService>.Instance,
+                new Microsoft.Extensions.Caching.Memory.MemoryCache(new Microsoft.Extensions.Caching.Memory.MemoryCacheOptions())),
+            maintenanceRepo: Mock.Of<IMaintenanceRepository>(),
+            offboardingRepo: repo,
+            offboardingEnqueuer: enqueuer,
+            previewWhitelistService: BuildPreviewWhitelistService(configRepo));
+        return (sut, repo, enqueuer, configRepo);
+    }
+
+    [Theory]
+    [InlineData(true)]
+    [InlineData(false)]
+    public async Task OffboardTenant_CarriesTheLiveTrialConsumedOntoThePointer(bool trialConsumed)
+    {
+        var live = TenantConfiguration.CreateDefault(TenantId);
+        live.DomainName = "contoso.com";
+        live.TrialConsumed = trialConsumed;
+        var (sut, repo, enqueuer, _) = BuildForOffboard(live);
+        var (req, context) = EndpointHarness.Request(TenantId);
+
+        var response = await sut.OffboardTenant(req, TenantId, context);
+
+        Assert.Equal(System.Net.HttpStatusCode.Accepted, response.StatusCode);
+        Assert.Equal(trialConsumed, repo.Pointers[TenantId].Pointer.TrialConsumed);
+        Assert.Single(enqueuer.Sent);
+    }
+
+    [Fact]
+    public async Task OffboardTenant_ConfigReadFails_RefusesBeforeAnyStateIsWritten()
+    {
+        // Proceeding without the read would reset the one-trial rule for the returning tenant.
+        var (sut, repo, enqueuer, configRepo) = BuildForOffboard(null, new RequestFailedException(503, "Server busy"));
+        var (req, context) = EndpointHarness.Request(TenantId);
+
+        var response = await sut.OffboardTenant(req, TenantId, context);
+
+        Assert.Equal(System.Net.HttpStatusCode.InternalServerError, response.StatusCode);
+        Assert.Empty(repo.History);
+        Assert.Empty(repo.Pointers);
+        Assert.Empty(repo.Markers);
+        Assert.Empty(enqueuer.Sent);
+        configRepo.VerifyNoTenantConfigWrite();
+    }
+
     [Fact]
     public async Task UpsertPointer_409OnInsert_FallsBackToEtagUpdate()
     {
@@ -273,7 +366,7 @@ public sealed class TenantOffboardFunctionTests
         var sut = Build(repoMock.Object);
         var newHistoryRowKey = $"{DateTime.UtcNow:yyyyMMddHHmmssfff}_{TenantId}";
 
-        await sut.UpsertPointerWithCasAsync(TenantId, newHistoryRowKey, DateTime.UtcNow);
+        await sut.UpsertPointerWithCasAsync(TenantId, newHistoryRowKey, DateTime.UtcNow, trialConsumed: false);
 
         Assert.NotNull(updatedPointer);
         Assert.Equal(newHistoryRowKey, updatedPointer!.LatestHistoryRowKey);
@@ -325,7 +418,7 @@ public sealed class TenantOffboardFunctionTests
         var sut = Build(repoMock.Object);
         var newHistoryRowKey = $"{DateTime.UtcNow:yyyyMMddHHmmssfff}_{TenantId}";
 
-        await sut.UpsertPointerWithCasAsync(TenantId, newHistoryRowKey, DateTime.UtcNow);
+        await sut.UpsertPointerWithCasAsync(TenantId, newHistoryRowKey, DateTime.UtcNow, trialConsumed: false);
 
         Assert.Equal(2, updateCalls);
         Assert.Equal(2, readIdx);
@@ -484,7 +577,7 @@ public sealed class TenantOffboardFunctionTests
         var historyRowKey = $"{DateTime.UtcNow:yyyyMMddHHmmssfff}_{TenantId}";
 
         await Assert.ThrowsAsync<InvalidOperationException>(() =>
-            sut.UpsertPointerWithCasAsync(TenantId, historyRowKey, DateTime.UtcNow));
+            sut.UpsertPointerWithCasAsync(TenantId, historyRowKey, DateTime.UtcNow, trialConsumed: false));
     }
 
     // ── Cache-Drain-Barrier (plan v2 §2.2 / §2.3) ─────────────────────────────
@@ -638,27 +731,42 @@ public sealed class TenantOffboardFunctionTests
         var (sut, configRepo, cache, _) = BuildWithConfigService();
         var existing = TenantConfiguration.CreateDefault(TenantId);
         existing.Disabled = false;
-        configRepo.Setup(r => r.GetTenantConfigurationAsync(TenantId)).ReturnsAsync(existing);
-
-        TenantConfiguration? saved = null;
-        configRepo.Setup(r => r.SaveTenantConfigurationAsync(It.IsAny<TenantConfiguration>(), It.IsAny<string?>(), It.IsAny<string?>()))
-            .Callback<TenantConfiguration, string?, string?>((c, _, _) => saved = c)
-            .ReturnsAsync(true);
+        existing.DomainName = "contoso.com";
+        var store = configRepo.BackWithRow(existing);
 
         // Seed cache with stale "Disabled=false" entry so we can prove invalidation.
         cache.Set($"tenant-config:{TenantId}", existing);
 
         await sut.EnsureTenantDisabledAsync(TenantId, "alice@contoso.com");
 
-        Assert.NotNull(saved);
-        Assert.True(saved!.Disabled);
+        var (saved, source, reason) = Assert.Single(store.Writes);
+        Assert.True(saved.Disabled);
         Assert.Equal("Offboarding in progress", saved.DisabledReason);
         Assert.Null(saved.DisabledUntil);
         Assert.Equal("alice@contoso.com", saved.UpdatedBy);
+        Assert.Equal("contoso.com", saved.DomainName); // the rest of the row survives
+        Assert.Equal(("offboard", "Disabled-gate before cascade"), (source, reason));
         Assert.False(cache.TryGetValue($"tenant-config:{TenantId}", out _),
             "cache entry must be invalidated so warm reads pick up the new Disabled=true on next access");
     }
 
+    [Fact]
+    public async Task EnsureTenantDisabled_DecidesOnTheStoredRow_NotAStaleCachedView()
+    {
+        // A cached copy still says "not disabled"; the stored row is already the tombstone.
+        var (sut, configRepo, cache, _) = BuildWithConfigService();
+        var tombstone = TenantConfiguration.CreateDefault(TenantId);
+        tombstone.Disabled = true;
+        tombstone.DisabledReason = "Offboarding in progress";
+        var store = configRepo.BackWithRow(tombstone);
+        cache.Set($"tenant-config:{TenantId}", TenantConfiguration.CreateDefault(TenantId));
+
+        var result = await sut.EnsureTenantDisabledAsync(TenantId, "alice@contoso.com");
+
+        Assert.Equal(TenantOffboardFunction.EnsureDisabledOutcome.AlreadyTombstone, result.Outcome);
+        Assert.Empty(store.Writes);
+    }
+
     [Fact]
     public async Task EnsureTenantDisabled_AlreadyDisabledWithSameReason_DoesNotRewrite()
     {
@@ -666,12 +774,12 @@ public sealed class TenantOffboardFunctionTests
         var existing = TenantConfiguration.CreateDefault(TenantId);
         existing.Disabled = true;
         existing.DisabledReason = "Offboarding in progress";
-        configRepo.Setup(r => r.GetTenantConfigurationAsync(TenantId)).ReturnsAsync(existing);
+        var store = configRepo.BackWithRow(existing);
 
         await sut.EnsureTenantDisabledAsync(TenantId, "alice@contoso.com");
 
-        // Idempotent: no Save thrash on resume-clicks while the marker is already in flight.
-        configRepo.Verify(r => r.SaveTenantConfigurationAsync(It.IsAny<TenantConfiguration>(), It.IsAny<string?>(), It.IsAny<string?>()), Times.Never);
+        // Idempotent: no write thrash on resume-clicks while the marker is already in flight.
+        Assert.Empty(store.Writes);
     }
 
     [Fact]
@@ -681,21 +789,19 @@ public sealed class TenantOffboardFunctionTests
         // skip the enqueue. We just assert the exception propagates — the caller's
         // try/catch + 500-return is exercised by integration tests.
         var (sut, configRepo, _, _) = BuildWithConfigService();
-        configRepo.Setup(r => r.GetTenantConfigurationAsync(TenantId))
-            .ThrowsAsync(new InvalidOperationException("storage unavailable"));
+        configRepo.BackWithRow(TenantConfiguration.CreateDefault(TenantId)).ThrowOnRead =
+            new InvalidOperationException("storage unavailable");
 
         await Assert.ThrowsAsync<InvalidOperationException>(() =>
             sut.EnsureTenantDisabledAsync(TenantId, "alice@contoso.com"));
     }
 
     [Fact]
-    public async Task EnsureTenantDisabled_StorageThrowsOnSave_PropagatesAsExpected()
+    public async Task EnsureTenantDisabled_StorageThrowsOnWrite_PropagatesAsExpected()
     {
         var (sut, configRepo, _, _) = BuildWithConfigService();
-        configRepo.Setup(r => r.GetTenantConfigurationAsync(TenantId))
-            .ReturnsAsync(TenantConfiguration.CreateDefault(TenantId));
-        configRepo.Setup(r => r.SaveTenantConfigurationAsync(It.IsAny<TenantConfiguration>(), It.IsAny<string?>(), It.IsAny<string?>()))
-            .ThrowsAsync(new InvalidOperationException("storage save failed"));
+        configRepo.BackWithRow(TenantConfiguration.CreateDefault(TenantId)).ThrowOnWrite =
+            new InvalidOperationException("storage write failed");
 
         await Assert.ThrowsAsync<InvalidOperationException>(() =>
             sut.EnsureTenantDisabledAsync(TenantId, "alice@contoso.com"));
@@ -709,35 +815,25 @@ public sealed class TenantOffboardFunctionTests
         // The fix creates a fresh CreateDefault row with Disabled=true so the auth-gate is
         // committed even in this edge case.
         var (sut, configRepo, _, _) = BuildWithConfigService();
-        configRepo.Setup(r => r.GetTenantConfigurationAsync(TenantId))
-            .ReturnsAsync((TenantConfiguration?)null);
+        var store = configRepo.BackWithRow(null);
 
-        TenantConfiguration? saved = null;
-        configRepo.Setup(r => r.SaveTenantConfigurationAsync(It.IsAny<TenantConfiguration>(), It.IsAny<string?>(), It.IsAny<string?>()))
-            .Callback<TenantConfiguration, string?, string?>((c, _, _) => saved = c)
-            .ReturnsAsync(true);
+        var result = await sut.EnsureTenantDisabledAsync(TenantId, "alice@contoso.com");
 
-        await sut.EnsureTenantDisabledAsync(TenantId, "alice@contoso.com");
-
-        Assert.NotNull(saved);
-        Assert.Equal(TenantId, saved!.TenantId);
+        var saved = Assert.Single(store.Writes).Row;
+        Assert.Equal(TenantId, saved.TenantId);
         Assert.True(saved.Disabled, "fresh tombstone must carry Disabled=true");
         Assert.Equal("Offboarding in progress", saved.DisabledReason);
         Assert.Equal("alice@contoso.com", saved.UpdatedBy);
+        Assert.Equal(TenantOffboardFunction.EnsureDisabledOutcome.WroteNewTombstone, result.Outcome);
     }
 
     [Fact]
-    public async Task EnsureTenantDisabled_SaveReturnsFalse_ThrowsAsHardFail()
+    public async Task EnsureTenantDisabled_EveryWriteLost_ThrowsAsHardFail()
     {
-        // PR3.B-revised Finding 2: TableConfigRepository.SaveTenantConfigurationAsync
-        // returns false on transient storage failure (it does NOT throw). Ignoring that
-        // return-value would let the caller proceed to enqueue without the auth-gate
-        // committed. Turn it into a hard exception so the offboard is NOT enqueued.
+        // Without Disabled=true the caller would proceed to enqueue without the auth-gate
+        // committed: every lost conditional write is a hard exception, never a silent skip.
         var (sut, configRepo, _, _) = BuildWithConfigService();
-        configRepo.Setup(r => r.GetTenantConfigurationAsync(TenantId))
-            .ReturnsAsync(TenantConfiguration.CreateDefault(TenantId));
-        configRepo.Setup(r => r.SaveTenantConfigurationAsync(It.IsAny<TenantConfiguration>(), It.IsAny<string?>(), It.IsAny<string?>()))
-            .ReturnsAsync(false);  // transient storage rejection — NOT a throw.
+        configRepo.BackWithRow(TenantConfiguration.CreateDefault(TenantId)).RejectWrites = true;
 
         var ex = await Assert.ThrowsAsync<InvalidOperationException>(() =>
             sut.EnsureTenantDisabledAsync(TenantId, "alice@contoso.com"));
@@ -745,15 +841,10 @@ public sealed class TenantOffboardFunctionTests
     }
 
     [Fact]
-    public async Task EnsureTenantDisabled_NoConfigRow_SaveReturnsFalse_ThrowsAsHardFail()
+    public async Task EnsureTenantDisabled_NoConfigRow_EveryWriteLost_ThrowsAsHardFail()
     {
-        // Combined edge case: missing row → we'd create a fresh tombstone → Save fails →
-        // hard fail (otherwise the caller would proceed without an auth-gate).
         var (sut, configRepo, _, _) = BuildWithConfigService();
-        configRepo.Setup(r => r.GetTenantConfigurationAsync(TenantId))
-            .ReturnsAsync((TenantConfiguration?)null);
-        configRepo.Setup(r => r.SaveTenantConfigurationAsync(It.IsAny<TenantConfiguration>(), It.IsAny<string?>(), It.IsAny<string?>()))
-            .ReturnsAsync(false);
+        configRepo.BackWithRow(null).RejectWrites = true;
 
         await Assert.ThrowsAsync<InvalidOperationException>(() =>
             sut.EnsureTenantDisabledAsync(TenantId, "alice@contoso.com"));
@@ -766,12 +857,12 @@ public sealed class TenantOffboardFunctionTests
         var tombstone = TenantConfiguration.CreateDefault(TenantId);
         tombstone.Disabled = true;
         tombstone.DisabledReason = "Offboarding in progress";
-        configRepo.Setup(r => r.GetTenantConfigurationAsync(TenantId)).ReturnsAsync(tombstone);
+        var store = configRepo.BackWithRow(tombstone);
 
         var result = await sut.EnsureTenantDisabledAsync(TenantId, "alice@contoso.com");
 
         Assert.Equal(TenantOffboardFunction.EnsureDisabledOutcome.AlreadyTombstone, result.Outcome);
-        configRepo.Verify(r => r.SaveTenantConfigurationAsync(It.IsAny<TenantConfiguration>(), It.IsAny<string?>(), It.IsAny<string?>()), Times.Never);
+        Assert.Empty(store.Writes);
     }
 
     [Fact]
@@ -780,13 +871,13 @@ public sealed class TenantOffboardFunctionTests
         var (sut, configRepo, _, _) = BuildWithConfigService();
         var existing = TenantConfiguration.CreateDefault(TenantId);
         existing.Disabled = false;
-        configRepo.Setup(r => r.GetTenantConfigurationAsync(TenantId)).ReturnsAsync(existing);
-        configRepo.Setup(r => r.SaveTenantConfigurationAsync(It.IsAny<TenantConfiguration>(), It.IsAny<string?>(), It.IsAny<string?>()))
-            .ReturnsAsync(true);
+        var store = configRepo.BackWithRow(existing);
 
         var result = await sut.EnsureTenantDisabledAsync(TenantId, "alice@contoso.com");
 
         Assert.Equal(TenantOffboardFunction.EnsureDisabledOutcome.WroteNewTombstone, result.Outcome);
+        // The anchor of the drain-deadline math is the stored write time.
+        Assert.Equal(Assert.Single(store.Writes).Row.LastUpdated, result.TombstoneLastUpdated);
     }
 
     // ── Resume-path cache-drain bypass fix (Codex review) ────────────────────
@@ -830,10 +921,7 @@ public sealed class TenantOffboardFunctionTests
         var configRepoMock = new Mock<IConfigRepository>();
         var notYetDisabled = TenantConfiguration.CreateDefault(TenantId);
         notYetDisabled.Disabled = false;
-        configRepoMock.Setup(r => r.GetTenantConfigurationAsync(It.IsAny<string>()))
-            .ReturnsAsync(notYetDisabled);
-        configRepoMock.Setup(r => r.SaveTenantConfigurationAsync(It.IsAny<TenantConfiguration>(), It.IsAny<string?>(), It.IsAny<string?>()))
-            .ReturnsAsync(true);
+        configRepoMock.BackWithRow(notYetDisabled);
 
         var cache = new MemoryCache(new MemoryCacheOptions());
         var tenantConfigService = new TenantConfigurationService(
@@ -925,8 +1013,7 @@ public sealed class TenantOffboardFunctionTests
         tombstone.Disabled = true;
         tombstone.DisabledReason = "Offboarding in progress";
         tombstone.LastUpdated = DateTime.UtcNow.AddSeconds(-1); // freshly written ~1s ago
-        configRepoMock.Setup(r => r.GetTenantConfigurationAsync(It.IsAny<string>()))
-            .ReturnsAsync(tombstone);
+        configRepoMock.BackWithRow(tombstone);
 
         var cache = new MemoryCache(new MemoryCacheOptions());
         var tenantConfigService = new TenantConfigurationService(
@@ -1000,8 +1087,7 @@ public sealed class TenantOffboardFunctionTests
         tombstone.Disabled = true;
         tombstone.DisabledReason = "Offboarding in progress";
         tombstone.LastUpdated = initiatedAt;
-        configRepoMock.Setup(r => r.GetTenantConfigurationAsync(It.IsAny<string>()))
-            .ReturnsAsync(tombstone);
+        configRepoMock.BackWithRow(tombstone);
 
         var cache = new MemoryCache(new MemoryCacheOptions());
         var tenantConfigService = new TenantConfigurationService(
@@ -1043,8 +1129,8 @@ public sealed class TenantOffboardFunctionTests
         var historyAfter = repo.History[historyRowKey];
         Assert.Equal(originalEarliestProcessingAt, historyAfter.EarliestProcessingAt);
 
-        // No Save call (the tombstone was already correct).
-        configRepoMock.Verify(r => r.SaveTenantConfigurationAsync(It.IsAny<TenantConfiguration>(), It.IsAny<string?>(), It.IsAny<string?>()), Times.Never);
+        // No write (the tombstone was already correct).
+        configRepoMock.VerifyNoTenantConfigWrite();
     }
 
     // ── RetryFailedOffboardingAsync (operator retry of a Failed offboarding) ──────────
@@ -1208,7 +1294,7 @@ public sealed class TenantOffboardFunctionTests
         tombstone.Disabled = true;
         tombstone.DisabledReason = "Offboarding in progress";
         tombstone.LastUpdated = DateTime.UtcNow.AddMinutes(-2);
-        configRepoMock.Setup(r => r.GetTenantConfigurationAsync(It.IsAny<string>())).ReturnsAsync(tombstone);
+        configRepoMock.BackWithRow(tombstone);
         var cache = new MemoryCache(new MemoryCacheOptions());
         var service = new TenantConfigurationService(configRepoMock.Object, NullLogger<TenantConfigurationService>.Instance, cache);
         var enqueuer = new RecordingEnqueuer();
@@ -1227,7 +1313,7 @@ public sealed class TenantOffboardFunctionTests
         var delay = Assert.Single(enqueuer.VisibilityDelays);
         Assert.NotNull(delay);
         Assert.InRange(delay!.Value, TimeSpan.FromMinutes(3.5), TimeSpan.FromMinutes(4.1));
-        configRepoMock.Verify(r => r.SaveTenantConfigurationAsync(It.IsAny<TenantConfiguration>(), It.IsAny<string>(), It.IsAny<string>()), Times.Never);
+        configRepoMock.VerifyNoTenantConfigWrite();
     }
 
     [Fact]

src/Backend/AutopilotMonitor.Functions.Tests/TypedContractCompletionWireParityTests.cs

@@ -191,6 +191,7 @@ public class TypedContractCompletionWireParityTests
                     new
                     {
                         appName = "Contoso App",
+                        source = "ime",
                         totalInstalls = 4,
                         succeeded = 2,
                         skipped = 1,
@@ -234,6 +235,7 @@ public class TypedContractCompletionWireParityTests
                     new AppMetricsAppGroup
                     {
                         AppName = "Contoso App",
+                        Source = AppInstallSources.Ime,
                         TotalInstalls = 4,
                         Succeeded = 2,
                         Skipped = 1,
@@ -593,6 +595,7 @@ public class TypedContractCompletionWireParityTests
         var item = new AppsListItem
         {
             AppName = "Contoso App",
+            Source = AppInstallSources.Ime,
             AppType = "Win32",
             TotalInstalls = 5,
             Succeeded = 3,
@@ -610,6 +613,7 @@ public class TypedContractCompletionWireParityTests
         var anonymousItem = new
         {
             appName = "Contoso App",
+            source = "ime",
             appType = "Win32",
             totalInstalls = 5,
             succeeded = 3,
@@ -677,6 +681,7 @@ public class TypedContractCompletionWireParityTests
             {
                 success = true,
                 appName = "Contoso App",
+                source = "ime",
                 appType = string.Empty,
                 windowDays = 30,
                 collisionExcluded = 0,
@@ -709,6 +714,7 @@ public class TypedContractCompletionWireParityTests
             {
                 Success = true,
                 AppName = "Contoso App",
+                Source = AppInstallSources.Ime,
                 AppType = string.Empty,
                 WindowDays = 30,
                 CollisionExcluded = 0,

src/Backend/AutopilotMonitor.Functions.Tests/TypedRequestGuardTests.cs

@@ -62,7 +62,6 @@ public class TypedRequestGuardTests
     {
         ["Functions/Config/PatchTenantConfigurationFieldsFunction.cs"] = 1,
         ["Functions/Config/PatchAdminConfigurationFunction.cs"] = 1,
-        ["Functions/Config/UpdateTenantConfigurationFunction.cs"] = 1,
         ["Functions/Rules/AnalyzeRulesFunction.cs"] = 2,
         ["Functions/Rules/DryRunAnalyzeRuleFunction.cs"] = 1,
         ["Functions/Rules/GatherRulesFunction.cs"] = 2,

src/Backend/AutopilotMonitor.Functions.Tests/WebhookConnectGateTests.cs

@@ -0,0 +1,247 @@
+using System.Net;
+using System.Net.Http;
+using System.Net.Security;
+using System.Net.Sockets;
+using System.Security.Cryptography;
+using System.Security.Cryptography.X509Certificates;
+using System.Text;
+using AutopilotMonitor.Functions.Helpers;
+using AutopilotMonitor.Functions.Security;
+using AutopilotMonitor.Functions.Services.Notifications;
+using AutopilotMonitor.Shared.Models.Notifications;
+using Microsoft.Extensions.DependencyInjection;
+using Microsoft.Extensions.Logging.Abstractions;
+
+namespace AutopilotMonitor.Functions.Tests;
+
+/// <summary>
+/// The webhook SSRF gate on real loopback sockets. DNS is injected: <c>rebind.example.test</c> does
+/// not exist, so a socket reaching a listener proves it used the gate's answer rather than a
+/// second resolution.
+/// </summary>
+public class WebhookConnectGateTests
+{
+    private const string Host = "rebind.example.test";
+
+    private sealed class CountingResolver
+    {
+        private readonly IPAddress[] _answer;
+        public List<string> Calls { get; } = new();
+        public CountingResolver(params string[] answer) => _answer = answer.Select(IPAddress.Parse).ToArray();
+
+        public Task<IPAddress[]> Resolve(string host, CancellationToken _)
+        {
+            Calls.Add(host);
+            return Task.FromResult(_answer);
+        }
+    }
+
+    private static HttpClient ClientThrough(Func<SocketsHttpConnectionContext, CancellationToken, ValueTask<Stream>> connect,
+        X509Certificate2? trustedServerCertificate = null)
+    {
+        var handler = new SocketsHttpHandler { AllowAutoRedirect = false, ConnectCallback = connect };
+        if (trustedServerCertificate != null)
+        {
+            // Pinned to the test's own self-signed certificate — validation stays on, only the trust anchor differs.
+            var pinned = trustedServerCertificate.GetCertHashString();
+            handler.SslOptions.RemoteCertificateValidationCallback = (_, cert, _, _) => cert?.GetCertHashString() == pinned;
+        }
+        // Short timeout: with a broken gate the TLS handshake waits on a listener that never accepts.
+        return new HttpClient(handler) { Timeout = TimeSpan.FromSeconds(10) };
+    }
+
+    private static TcpListener StartListener()
+    {
+        var listener = new TcpListener(IPAddress.Loopback, 0);
+        listener.Start();
+        return listener;
+    }
+
+    private static int PortOf(TcpListener listener) => ((IPEndPoint)listener.LocalEndpoint).Port;
+
+    [Fact]
+    public async Task PrivateAnswer_FailsClosed_WithoutOpeningASocket()
+    {
+        var listener = StartListener();
+        try
+        {
+            var resolver = new CountingResolver("127.0.0.1");
+            using var client = ClientThrough(SsrfGuard.CreateConnectCallback(resolver.Resolve));
+
+            var ex = await Assert.ThrowsAsync<HttpRequestException>(() => client.GetAsync($"https://{Host}:{PortOf(listener)}/hook"));
+
+            var refused = Assert.IsType<SsrfException>(ex.InnerException);
+            Assert.Contains("private or reserved", refused.Message);
+            Assert.Equal(new[] { Host }, resolver.Calls);
+            Assert.False(listener.Pending());
+        }
+        finally
+        {
+            listener.Stop();
+        }
+    }
+
+    [Fact]
+    public async Task MixedAnswer_WithOnePrivateAddress_FailsClosed()
+    {
+        var resolver = new CountingResolver("1.1.1.1", "169.254.169.254");
+        using var client = ClientThrough(SsrfGuard.CreateConnectCallback(resolver.Resolve));
+
+        var ex = await Assert.ThrowsAsync<HttpRequestException>(() => client.GetAsync($"https://{Host}/hook"));
+
+        Assert.IsType<SsrfException>(ex.InnerException);
+    }
+
+    [Fact]
+    public async Task PrivateIpLiteral_FailsClosed_WithoutResolving()
+    {
+        var listener = StartListener();
+        try
+        {
+            var resolver = new CountingResolver("1.1.1.1");
+            using var client = ClientThrough(SsrfGuard.CreateConnectCallback(resolver.Resolve));
+
+            var ex = await Assert.ThrowsAsync<HttpRequestException>(() => client.GetAsync($"https://127.0.0.1:{PortOf(listener)}/hook"));
+
+            Assert.IsType<SsrfException>(ex.InnerException);
+            Assert.Empty(resolver.Calls);
+            Assert.False(listener.Pending());
+        }
+        finally
+        {
+            listener.Stop();
+        }
+    }
+
+    [Fact]
+    public async Task PlainHttp_IsRefused_BeforeResolving()
+    {
+        var resolver = new CountingResolver("1.1.1.1");
+        using var client = ClientThrough(SsrfGuard.CreateConnectCallback(resolver.Resolve));
+
+        var ex = await Assert.ThrowsAsync<HttpRequestException>(() => client.GetAsync($"http://{Host}/hook"));
+
+        Assert.Contains("HTTPS", Assert.IsType<SsrfException>(ex.InnerException).Message);
+        Assert.Empty(resolver.Calls);
+    }
+
+    [Fact]
+    public async Task UnresolvableOrEmptyAnswer_FailsClosed()
+    {
+        Task<IPAddress[]> NotFound(string _, CancellationToken __) => throw new SocketException((int)SocketError.HostNotFound);
+        using (var client = ClientThrough(SsrfGuard.CreateConnectCallback(NotFound)))
+        {
+            var ex = await Assert.ThrowsAsync<HttpRequestException>(() => client.GetAsync($"https://{Host}/hook"));
+            Assert.Contains("resolve", Assert.IsType<SsrfException>(ex.InnerException).Message);
+        }
+
+        using (var client = ClientThrough(SsrfGuard.CreateConnectCallback(new CountingResolver().Resolve)))
+        {
+            var ex = await Assert.ThrowsAsync<HttpRequestException>(() => client.GetAsync($"https://{Host}/hook"));
+            Assert.Contains("no addresses", Assert.IsType<SsrfException>(ex.InnerException).Message);
+        }
+    }
+
+    [Fact]
+    public async Task AllowedAnswer_ConnectsToExactlyThatAddress_AndTlsStillTargetsTheRequestHost()
+    {
+        using var certificate = CreateServerCertificate(Host);
+        var listener = StartListener();
+        try
+        {
+            string? serverName = null;
+            var server = Task.Run(async () =>
+            {
+                using var tcp = await listener.AcceptTcpClientAsync();
+                using var tls = new SslStream(tcp.GetStream());
+                await tls.AuthenticateAsServerAsync(new SslServerAuthenticationOptions { ServerCertificate = certificate });
+                serverName = tls.TargetHostName;
+                var buffer = new byte[4096];
+                var request = new StringBuilder();
+                while (!request.ToString().Contains("\r\n\r\n"))
+                {
+                    var read = await tls.ReadAsync(buffer);
+                    if (read == 0) break;
+                    request.Append(Encoding.ASCII.GetString(buffer, 0, read));
+                }
+                await tls.WriteAsync(Encoding.ASCII.GetBytes("HTTP/1.1 200 OK\r\nContent-Length: 2\r\nConnection: close\r\n\r\nok"));
+            });
+
+            // The classifier is widened to reach the loopback listener; the socket path is the production one.
+            var resolver = new CountingResolver("127.0.0.1");
+            using var client = ClientThrough(SsrfGuard.CreateConnectCallback(resolver.Resolve, _ => false), certificate);
+
+            var response = await client.GetAsync($"https://{Host}:{PortOf(listener)}/hook");
+            await server;
+
+            Assert.Equal(HttpStatusCode.OK, response.StatusCode);
+            Assert.Equal("ok", await response.Content.ReadAsStringAsync());
+            Assert.Equal(Host, serverName);
+            Assert.Equal(new[] { Host }, resolver.Calls);
+        }
+        finally
+        {
+            listener.Stop();
+        }
+    }
+
+    [Fact]
+    public void WebhookClient_PrimaryHandler_ConnectsOnlyThroughTheGate()
+    {
+        var services = new ServiceCollection();
+        services.AddLogging();
+        services.AddSingleton<ResiliencePolicies>();
+        services.AddWebhookNotificationHttpClient();
+        using var provider = services.BuildServiceProvider();
+
+        HttpMessageHandler handler = provider.GetRequiredService<IHttpMessageHandlerFactory>().CreateHandler(nameof(WebhookNotificationService));
+        while (handler is DelegatingHandler delegating)
+            handler = delegating.InnerHandler!;
+
+        var primary = Assert.IsType<SocketsHttpHandler>(handler);
+        Assert.Same(SsrfGuard.ConnectToPublicAddress, primary.ConnectCallback);
+        Assert.False(primary.AllowAutoRedirect);
+    }
+
+    [Fact]
+    public async Task NotificationPolicy_DoesNotRetryARefusedDestination()
+    {
+        var policies = new ResiliencePolicies(NullLoggerFactory.Instance);
+        var attempts = 0;
+
+        await Assert.ThrowsAsync<HttpRequestException>(() => policies.Notification.ExecuteAsync(() =>
+        {
+            attempts++;
+            throw new HttpRequestException("refused", new SsrfException("Webhook URL targets a private or reserved network."));
+        }));
+
+        Assert.Equal(1, attempts);
+    }
+
+    [Fact]
+    public async Task TestNotification_ReportsTheRefusal()
+    {
+        var resolver = new CountingResolver("10.0.0.8");
+        var webhook = new WebhookNotificationService(ClientThrough(SsrfGuard.CreateConnectCallback(resolver.Resolve)),
+            NullLogger<WebhookNotificationService>.Instance);
+
+        var result = await webhook.SendNotificationWithResultAsync($"https://{Host}/hook", WebhookProviderType.GenericJson,
+            new NotificationAlert { Title = "T", Summary = "S" });
+
+        Assert.False(result.Success);
+        Assert.Equal("Webhook URL targets a private or reserved network.", result.Message);
+    }
+
+    private static X509Certificate2 CreateServerCertificate(string host)
+    {
+        using var key = ECDsa.Create(ECCurve.NamedCurves.nistP256);
+        var request = new CertificateRequest($"CN={host}", key, HashAlgorithmName.SHA256);
+        var san = new SubjectAlternativeNameBuilder();
+        san.AddDnsName(host);
+        request.CertificateExtensions.Add(san.Build());
+        request.CertificateExtensions.Add(new X509EnhancedKeyUsageExtension(new OidCollection { new Oid("1.3.6.1.5.5.7.3.1") }, false));
+        using var ephemeral = request.CreateSelfSigned(DateTimeOffset.UtcNow.AddMinutes(-5), DateTimeOffset.UtcNow.AddHours(1));
+        // Round-trip through PKCS#12: SChannel cannot serve a certificate whose key is ephemeral.
+        return X509CertificateLoader.LoadPkcs12(ephemeral.Export(X509ContentType.Pfx), null);
+    }
+}

src/Backend/AutopilotMonitor.Functions/DataAccess/TableStorage/PerPartitionFanOutMerge.cs

@@ -3,159 +3,158 @@ using System.Collections.Generic;
 using System.Linq;
 using System.Text;
 using System.Text.Json;
+using System.Threading;
+using System.Threading.Tasks;
+using AutopilotMonitor.Functions.Helpers;
+using AutopilotMonitor.Shared.Pagination;
 
 namespace AutopilotMonitor.Functions.DataAccess.TableStorage
 {
     /// <summary>
-    /// Generic helpers driving cross-partition merge-pagination for global views
-    /// where Azure-Tables' native (PK asc, RK asc) ordering would otherwise
-    /// surface partitions one-at-a-time instead of globally newest-first.
-    /// Used by ops-events (categories as partitions) and global audit logs
-    /// (tenants as partitions).
+    /// Newest-first paging across the partitions of one table: global ops events (categories as
+    /// partitions) and the global audit log (tenants as partitions). Azure Tables pages a
+    /// cross-partition query in (PK asc, RK asc) order, which would surface one partition at a time,
+    /// so every page runs one PartitionKey query per partition and merges the results.
     /// </summary>
     /// <remarks>
     /// <para>
-    /// Per-partition continuation anchored on RowKey (not Azure's opaque token):
-    /// <c>LastRowKey</c> = the oldest RowKey we have already returned for this
-    /// partition. Next fetch filters <c>RowKey gt LastRowKey</c> — natural
-    /// ordering within the partition (revtick → newest-first) means this skips
-    /// exactly what we showed and re-considers everything older.
+    /// Invariant: every partition uses the same reverse-tick RowKey scheme, so the global order is
+    /// (RowKey asc, Partition asc) and that order is the display order. The merge sorts by exactly
+    /// that key and never by a resolved timestamp column — a merge key that disagrees with the
+    /// RowKey inside one partition would let the watermark skip rows.
     /// </para>
     /// <para>
-    /// Why not Azure's continuation token? Cross-partition merge can drop items
-    /// that were fetched but didn't survive the top-pageSize cut. If we advanced
-    /// an Azure cursor past those leftovers, we'd lose them forever. With a
-    /// RowKey bound we simply don't advance the continuation for partitions whose
-    /// fetched items lost the merge — they get re-fetched next call and compete
-    /// again at a lower floor.
+    /// The continuation is the (RowKey, Partition) of the last returned row, so its size does not
+    /// depend on the number of partitions. The earlier per-partition continuation map kept one entry
+    /// for every partition with rows and outgrew the host's URL limit (HTTP 414) on the tenant
+    /// fan-out.
     /// </para>
     /// <para>
-    /// Continuation shape: only <em>active</em> partitions are encoded. Missing
-    /// from the map = exhausted (or never had any data). For 200-tenant audit
-    /// fan-outs that previously kept exhausted entries around, the encoded
-    /// token would balloon past 30 KB and trip URL-length limits before
-    /// reaching the function (browser <c>net::ERR_FAILED</c> on Next click).
-    /// Now the token shrinks page-over-page as partitions exhaust.
+    /// Each partition reads at most <c>pageSize + 1</c> rows after the watermark. More rows exist
+    /// exactly when more than <c>pageSize</c> were read in total: if no partition reached its limit,
+    /// every partition is exhausted.
     /// </para>
     /// </remarks>
     internal static class PerPartitionFanOutMerge
     {
-        internal sealed record PartitionContinuation(string? LastRowKey);
+        /// <summary>The (RowKey, Partition) of the last row a page returned.</summary>
+        internal sealed record Watermark(string RowKey, string Partition);
 
-        internal sealed record PartitionFetchResult<T>(
-            string Partition,
-            IReadOnlyList<(string RowKey, T Item)> Items);
-
-        /// <summary>
-        /// Pure merge-and-advance step. Given pre-fetched per-partition items
-        /// (already newest-first within each list) and the previous per-partition
-        /// continuation map, returns the top-pageSize merged items plus the
-        /// next continuation map. Partitions with nothing left to return are
-        /// dropped from the map entirely — consumers treat "missing" as
-        /// exhausted on subsequent calls (see class remarks).
-        /// </summary>
-        public static (List<T> Items, Dictionary<string, PartitionContinuation> NextContinuations)
-            MergeAndAdvance<T>(
-                IReadOnlyList<PartitionFetchResult<T>> fetched,
-                IReadOnlyDictionary<string, PartitionContinuation> priorContinuations,
-                int pageSize,
-                Func<T, DateTime> timestampSelector)
+        /// <summary>Lower RowKey bound of one partition's read.</summary>
+        internal sealed record RowKeyBound(string RowKey, bool Inclusive)
         {
-            var allFetched = fetched
-                .SelectMany(r => r.Items.Select(t => (part: r.Partition, rk: t.RowKey, item: t.Item)))
-                .OrderByDescending(t => timestampSelector(t.item))
-                .ToList();
-
-            var merged = allFetched.Take(pageSize).ToList();
-            var returnedItems = merged.Select(t => t.item).ToList();
-
-            var newContinuations = new Dictionary<string, PartitionContinuation>(StringComparer.Ordinal);
-            foreach (var r in fetched)
-            {
-                var returnedFromPart = merged.Where(m => m.part == r.Partition).ToList();
-                if (returnedFromPart.Count > 0)
-                {
-                    // Advance to the oldest RowKey we returned for this partition.
-                    // Revtick scheme: larger RowKey string = older timestamp, so
-                    // OrdinalCompare-max gives us the oldest of the returned set.
-                    var oldestRk = returnedFromPart
-                        .Select(m => m.rk)
-                        .OrderBy(rk => rk, StringComparer.Ordinal)
-                        .Last();
-                    newContinuations[r.Partition] = new PartitionContinuation(oldestRk);
-                }
-                else if (r.Items.Count == 0)
-                {
-                    // No items returned from this partition's query → exhausted.
-                    // Drop from map entirely; the consumer's "missing = exhausted"
-                    // rule (when continuation was provided) does the right thing
-                    // and keeps the encoded token small.
-                }
-                else
-                {
-                    // Fetched some but they all lost the merge cut to other partitions.
-                    // Keep continuation where it was — they'll be re-fetched and likely
-                    // win on the next page when the floor drops. Wasted bandwidth, but
-                    // correct (no item silently dropped).
-                    priorContinuations.TryGetValue(r.Partition, out var prior);
-                    newContinuations[r.Partition] = prior ?? new PartitionContinuation(null);
-                }
-            }
-
-            return (returnedItems, newContinuations);
+            public string ToODataClause() => $"RowKey {(Inclusive ? "ge" : "gt")} '{RowKey.Replace("'", "''")}'";
         }
 
-        // Wire format v1 (legacy): every partition encoded with `{rk, x:bool}`.
-        // Wire format v2 (current): only ACTIVE partitions, `{rk}` only. Decoder
-        // accepts both and silently drops v1 `x:true` entries so old tokens
-        // emitted by previous deploys still page through cleanly.
-        private const int WireFormatVersion = 2;
+        /// <summary>
+        /// Delegate that reads up to <c>limit</c> rows of one partition, newest first (RowKey
+        /// ascending), restricted by <c>bound</c> when it is set. It must return every matching row
+        /// up to the limit — a short read is taken as "partition exhausted".
+        /// </summary>
+        internal delegate Task<List<(string RowKey, T Item)>> PartitionReader<T>(
+            string partition, RowKeyBound? bound, int limit, CancellationToken cancellationToken);
 
-        public static string EncodeMultiContinuation(IReadOnlyDictionary<string, PartitionContinuation> continuations)
+        /// <summary>
+        /// The rows of <paramref name="partition"/> that come strictly after the watermark in
+        /// (RowKey, Partition) order. A partition that sorts after the watermark's partition may still
+        /// hold rows with the watermark's own RowKey; one that sorts before (or is the same) may not.
+        /// </summary>
+        internal static RowKeyBound? BoundAfter(Watermark? watermark, string partition)
         {
-            var doc = new Dictionary<string, object?>
+            if (watermark == null) return null;
+            var inclusive = string.CompareOrdinal(partition, watermark.Partition) > 0;
+            return new RowKeyBound(watermark.RowKey, inclusive);
+        }
+
+        public static async Task<RawPage<T>> FetchPageAsync<T>(
+            IEnumerable<string> partitions,
+            int pageSize,
+            string? continuation,
+            PartitionReader<T> readPartition,
+            CancellationToken cancellationToken = default)
+        {
+            if (pageSize < 1) throw new ArgumentOutOfRangeException(nameof(pageSize));
+
+            Watermark? watermark = null;
+            if (!string.IsNullOrEmpty(continuation))
+            {
+                watermark = DecodeWatermark(continuation);
+                // The HMAC wrapper already rejected foreign tokens; one that passes it and still does
+                // not decode was minted by an older deploy. End the list rather than restart it.
+                if (watermark == null) return RawPage<T>.Empty;
+            }
+
+            var limit = pageSize + 1;
+            var rows = await BoundedFanOut.RunAsync(partitions, BoundedFanOut.CrossTenantConcurrency, async (partition, ct) =>
+            {
+                var read = await readPartition(partition, BoundAfter(watermark, partition), limit, ct).ConfigureAwait(false);
+                return read.Select(r => (Partition: partition, r.RowKey, r.Item)).ToList();
+            }, cancellationToken).ConfigureAwait(false);
+
+            return Merge(rows, pageSize);
+        }
+
+        /// <summary>
+        /// Pure merge step: orders the rows read from all partitions by (RowKey, Partition), returns
+        /// the first <paramref name="pageSize"/> and a continuation when rows are left over.
+        /// </summary>
+        internal static RawPage<T> Merge<T>(IEnumerable<(string Partition, string RowKey, T Item)> rows, int pageSize)
+        {
+            var ordered = rows
+                .OrderBy(r => r.RowKey, StringComparer.Ordinal)
+                .ThenBy(r => r.Partition, StringComparer.Ordinal)
+                .ToList();
+            var page = ordered.Take(pageSize).ToList();
+
+            string? next = null;
+            if (ordered.Count > pageSize)
+            {
+                var last = page[page.Count - 1];
+                next = EncodeWatermark(new Watermark(last.RowKey, last.Partition));
+            }
+            return new RawPage<T>(page.Select(r => r.Item).ToList(), next);
+        }
+
+        private const int WireFormatVersion = 3;
+
+        public static string EncodeWatermark(Watermark watermark)
+        {
+            var json = JsonSerializer.Serialize(new Dictionary<string, object>
             {
                 ["v"] = WireFormatVersion,
-                ["c"] = continuations.ToDictionary(
-                    kv => kv.Key,
-                    kv => (object?)new Dictionary<string, object?>
-                    {
-                        ["rk"] = kv.Value.LastRowKey,
-                    }),
-            };
-            var json = JsonSerializer.Serialize(doc);
+                ["rk"] = watermark.RowKey,
+                ["p"] = watermark.Partition,
+            });
             return Convert.ToBase64String(Encoding.UTF8.GetBytes(json));
         }
 
-        public static Dictionary<string, PartitionContinuation> DecodeMultiContinuation(string? raw)
+        /// <summary>Decodes a watermark; null for anything that is not a v3 watermark.</summary>
+        public static Watermark? DecodeWatermark(string? raw)
         {
-            var result = new Dictionary<string, PartitionContinuation>(StringComparer.Ordinal);
-            if (string.IsNullOrEmpty(raw)) return result;
-
+            if (string.IsNullOrEmpty(raw)) return null;
             try
             {
-                var bytes = Convert.FromBase64String(raw);
-                using var doc = JsonDocument.Parse(bytes);
-                if (!doc.RootElement.TryGetProperty("c", out var entries)) return result;
-                foreach (var prop in entries.EnumerateObject())
-                {
-                    // v1 back-compat: skip entries explicitly marked exhausted.
-                    // v2 never writes them, but old in-flight tokens may.
-                    if (prop.Value.TryGetProperty("x", out var xEl) && xEl.ValueKind == JsonValueKind.True)
-                        continue;
+                using var doc = JsonDocument.Parse(Convert.FromBase64String(raw));
+                var root = doc.RootElement;
+                if (root.ValueKind != JsonValueKind.Object) return null;
+                if (!root.TryGetProperty("v", out var v) || v.ValueKind != JsonValueKind.Number
+                    || !v.TryGetInt32(out var version) || version != WireFormatVersion)
+                    return null;
+                if (!root.TryGetProperty("rk", out var rk) || rk.ValueKind != JsonValueKind.String) return null;
+                if (!root.TryGetProperty("p", out var p) || p.ValueKind != JsonValueKind.String) return null;
 
-                    string? rk = prop.Value.TryGetProperty("rk", out var rkEl) && rkEl.ValueKind == JsonValueKind.String
-                        ? rkEl.GetString()
-                        : null;
-                    result[prop.Name] = new PartitionContinuation(rk);
-                }
+                var rowKey = rk.GetString();
+                var partition = p.GetString();
+                return string.IsNullOrEmpty(rowKey) || partition == null ? null : new Watermark(rowKey, partition);
             }
-            catch
+            catch (FormatException)
             {
-                // Malformed continuation → treat as empty (start over). The wire-
-                // layer fingerprint already rejects tampered tokens before reaching here.
+                return null;
+            }
+            catch (JsonException)
+            {
+                return null;
             }
-            return result;
         }
     }
 }

src/Backend/AutopilotMonitor.Functions/DataAccess/TableStorage/TableAdminRepository.cs

@@ -498,7 +498,7 @@ namespace AutopilotMonitor.Functions.DataAccess.TableStorage
             return groupIds;
         }
 
-        public async Task<bool> RenameTenantGroupAsync(string groupId, string name)
+        public async Task<bool> UpdateTenantGroupAsync(string groupId, string? name, string? customerLabel)
         {
             if (string.IsNullOrWhiteSpace(groupId))
                 return false;
@@ -510,7 +510,10 @@ namespace AutopilotMonitor.Functions.DataAccess.TableStorage
                 var entity = result.Value;
                 if (entity == null) return false;
 
-                entity.Name = name?.Trim() ?? string.Empty;
+                if (name != null)
+                    entity.Name = name.Trim();
+                if (customerLabel != null)
+                    entity.CustomerLabel = customerLabel.Trim();
                 await _tenantGroupsTableClient.UpdateEntityAsync(entity, ETag.All);
                 return true;
             }
@@ -1022,6 +1025,7 @@ namespace AutopilotMonitor.Functions.DataAccess.TableStorage
                 group.CreatedBy = entity.CreatedBy;
                 group.CreatedAt = entity.CreatedDate ?? default;
                 group.OwnerTenantId = string.IsNullOrEmpty(entity.OwnerTenantId) ? null : entity.OwnerTenantId;
+                group.CustomerLabel = string.IsNullOrEmpty(entity.CustomerLabel) ? null : entity.CustomerLabel;
                 return true;
             }
 

src/Backend/AutopilotMonitor.Functions/DataAccess/TableStorage/TableConfigRepository.cs

@@ -78,27 +78,16 @@ namespace AutopilotMonitor.Functions.DataAccess.TableStorage
             }
         }
 
-        public Task<bool> SaveTenantConfigurationAsync(TenantConfiguration config)
-            => SaveTenantConfigurationAsync(config, backupSource: null, backupReason: null);
-
-        public async Task<bool> SaveTenantConfigurationAsync(
-            TenantConfiguration config, string? backupSource, string? backupReason)
+        public async Task<bool> TryCreateTenantConfigurationAsync(TenantConfiguration config)
         {
-            await TrySnapshotBeforeSaveAsync(
-                _tenantConfigTableClient, config.TenantId, "config",
-                ConvertFromTenantTableEntity, config, TenantBackupNoiseProperties,
-                config.UpdatedBy, backupSource, backupReason);
-
             try
             {
-                var entity = ConvertToTenantTableEntity(config);
-                await _tenantConfigTableClient.UpsertEntityAsync(entity, TableUpdateMode.Replace);
+                await _tenantConfigTableClient.AddEntityAsync(ConvertToTenantTableEntity(config));
                 return true;
             }
-            catch (Exception ex)
+            catch (RequestFailedException ex) when (ex.Status == 409)
             {
-                _logger.LogError(ex, "Error saving tenant configuration for {TenantId}", config.TenantId);
-                return false;
+                return false; // created concurrently — the caller re-reads and updates
             }
         }
 
@@ -134,6 +123,17 @@ namespace AutopilotMonitor.Functions.DataAccess.TableStorage
             }
         }
 
+        public async Task<bool> TryReplaceTenantConfigurationAsync(
+            TenantConfiguration config, string etag, string? backupSource, string? backupReason)
+        {
+            await TrySnapshotBeforeSaveAsync(
+                _tenantConfigTableClient, config.TenantId, "config",
+                ConvertFromTenantTableEntity, config, TenantBackupNoiseProperties,
+                config.UpdatedBy, backupSource, backupReason);
+
+            return await TryReplaceTenantConfigurationAsync(config, etag);
+        }
+
         /// <summary>
         /// Pre-write snapshot hook shared by the tenant- and admin-config save paths.
         /// Fail-SOFT by design: this is a safety net around long-standing writers
@@ -949,7 +949,7 @@ namespace AutopilotMonitor.Functions.DataAccess.TableStorage
                 EnrollmentSummaryTimeoutSeconds = entity.GetInt32("EnrollmentSummaryTimeoutSeconds"),
                 EnrollmentSummaryBrandingImageUrl = entity.GetString("EnrollmentSummaryBrandingImageUrl"),
                 EnrollmentSummaryLaunchRetrySeconds = entity.GetInt32("EnrollmentSummaryLaunchRetrySeconds"),
-                HelloWaitTimeoutSeconds = entity.GetInt32("HelloWaitTimeoutSeconds") ?? 30,
+                HelloWaitTimeoutSeconds = entity.GetInt32("HelloWaitTimeoutSeconds") ?? HelloWaitTimeout.DefaultSeconds,
                 AgentMaxLifetimeMinutes = entity.GetInt32("AgentMaxLifetimeMinutes"),
                 SendTraceEvents = entity.GetBoolean("SendTraceEvents") ?? true,
                 EnableLocalAdminAnalyzer = entity.GetBoolean("EnableLocalAdminAnalyzer"),

src/Backend/AutopilotMonitor.Functions/DataAccess/TableStorage/TableHardwareRejectionNotificationTracker.cs

@@ -18,6 +18,7 @@ namespace AutopilotMonitor.Functions.DataAccess.TableStorage
     ///   - F3 rule regression: "ruleregression|{ruleId-lower}" (trimmed; payload-carrying rows
     ///     that are refreshed while the episode is active and deleted on re-arm)
     ///   - app-version duration regression: "appversionregression|{app-lower}|{version-lower}"
+    ///     (non-IME channels: "appversionregression|{source}:{app-lower}|{version-lower}")
     ///     (trimmed + table-key-sanitized; same payload-carrying episode semantics — raw
     ///     AppName/CurrentVersion live in columns, the mapper never parses the RowKey)
     /// Race-safe via AddEntityAsync: Azure Table Storage returns 409 Conflict if the entity already exists.
@@ -309,7 +310,7 @@ namespace AutopilotMonitor.Functions.DataAccess.TableStorage
             }
         }
 
-        public async Task DeleteAppVersionRegressionAsync(string tenantId, string appName, string currentVersion)
+        public async Task DeleteAppVersionRegressionAsync(string tenantId, string source, string appName, string currentVersion)
         {
             if (string.IsNullOrWhiteSpace(tenantId)
                 || string.IsNullOrWhiteSpace(appName)
@@ -320,7 +321,7 @@ namespace AutopilotMonitor.Functions.DataAccess.TableStorage
             try
             {
                 await _table.DeleteEntityAsync(
-                    tenantId.ToLowerInvariant(), BuildAppVersionRegressionRowKey(appName, currentVersion));
+                    tenantId.ToLowerInvariant(), BuildAppVersionRegressionRowKey(source, appName, currentVersion));
             }
             catch (RequestFailedException ex) when (ex.Status == 404)
             {
@@ -576,10 +577,14 @@ namespace AutopilotMonitor.Functions.DataAccess.TableStorage
         /// two apps share an episode row — tolerable); the raw values live in columns and the
         /// mapper reads them, never the RowKey.
         /// </summary>
-        internal static string BuildAppVersionRegressionRowKey(string appName, string version)
+        internal static string BuildAppVersionRegressionRowKey(string? source, string appName, string version)
         {
             var app = SanitizeTableKey((appName ?? string.Empty).Trim().ToLowerInvariant());
             var ver = SanitizeTableKey((version ?? string.Empty).Trim().ToLowerInvariant());
+            // IME keys predate channels and keep their shape; other channels prefix the app part.
+            var channel = AppInstallSources.Normalize(source);
+            if (channel != AppInstallSources.Ime)
+                app = $"{SanitizeTableKey(channel)}:{app}";
             return $"{AppVersionRegressionRowKeyPrefix}{app}|{ver}";
         }
 
@@ -598,10 +603,11 @@ namespace AutopilotMonitor.Functions.DataAccess.TableStorage
         internal static TableEntity BuildAppVersionRegressionEntity(string tenantId, AppVersionRegressionAlert alert)
         {
             return new TableEntity(
-                tenantId.ToLowerInvariant(), BuildAppVersionRegressionRowKey(alert.AppName, alert.CurrentVersion))
+                tenantId.ToLowerInvariant(), BuildAppVersionRegressionRowKey(alert.Source, alert.AppName, alert.CurrentVersion))
             {
                 ["TenantId"] = tenantId,
                 ["AppName"] = alert.AppName,
+                ["Source"] = AppInstallSources.Normalize(alert.Source),
                 ["CurrentVersion"] = alert.CurrentVersion,
                 ["PreviousVersion"] = alert.PreviousVersion ?? string.Empty,
                 ["CurrentMedianSeconds"] = alert.CurrentMedianSeconds,
@@ -620,6 +626,8 @@ namespace AutopilotMonitor.Functions.DataAccess.TableStorage
             {
                 TenantId = entity.GetString("TenantId") ?? entity.PartitionKey,
                 AppName = entity.GetString("AppName") ?? string.Empty,
+                // Rows written before channels existed are IME episodes.
+                Source = AppInstallSources.Normalize(entity.GetString("Source")),
                 CurrentVersion = entity.GetString("CurrentVersion") ?? string.Empty,
                 PreviousVersion = entity.GetString("PreviousVersion") ?? string.Empty,
                 CurrentMedianSeconds = entity.GetInt32("CurrentMedianSeconds") ?? 0,

src/Backend/AutopilotMonitor.Functions/DataAccess/TableStorage/TableOffboardingAuditRepository.cs

@@ -282,6 +282,7 @@ namespace AutopilotMonitor.Functions.DataAccess.TableStorage
             ["LatestStatus"] = p.LatestStatus,
             ["LatestUpdatedAt"] = p.LatestUpdatedAt,
             ["OffboardCount"] = p.OffboardCount,
+            ["TrialConsumed"] = p.TrialConsumed,
         };
 
         private static OffboardingByTenantPointer MapByTenantPointer(TableEntity e) => new()
@@ -293,6 +294,8 @@ namespace AutopilotMonitor.Functions.DataAccess.TableStorage
             LatestStatus = e.GetString("LatestStatus") ?? "Initiated",
             LatestUpdatedAt = e.GetDateTime("LatestUpdatedAt") ?? default,
             OffboardCount = e.GetInt32("OffboardCount") ?? 1,
+            // Absent on pointers from before the column; false is correct for all of them (D-288).
+            TrialConsumed = e.GetBoolean("TrialConsumed") ?? false,
         };
     }
 }

src/Backend/AutopilotMonitor.Functions/DataAccess/TableStorage/TableOpsEventRepository.cs

@@ -115,60 +115,35 @@ namespace AutopilotMonitor.Functions.DataAccess.TableStorage
         // partitions would just trade certainty for cost — the coverage test guards the list instead.
         internal static readonly string[] AllCategories = OpsEventCategory.All;
 
-        private async Task<RawPage<OpsEventEntry>> FanOutAcrossCategoriesAsync(
+        private Task<RawPage<OpsEventEntry>> FanOutAcrossCategoriesAsync(
             DateTime? dateFrom, DateTime? dateTo, int pageSize, string? continuation,
             OpsEventQueryFilters? filters = null)
         {
-            var continuations = PerPartitionFanOutMerge.DecodeMultiContinuation(continuation);
-            // First-page request fans out across every category; subsequent
-            // pages restrict to whatever is still active in the continuation
-            // map. With only 6 fixed categories this never grew to a problem
-            // size, but the convention now matches the audit fan-out so the
-            // shared fan-out helper has one rule.
-            var isFirstPage = string.IsNullOrEmpty(continuation);
-
-            var activeCats = AllCategories
-                .Where(cat => isFirstPage || continuations.ContainsKey(cat))
-                .ToList();
-            if (activeCats.Count == 0)
-                return new RawPage<OpsEventEntry>(new List<OpsEventEntry>(), null);
-
-            var fetchTasks = activeCats.Select(async cat =>
-            {
-                continuations.TryGetValue(cat, out var catContinuation);
-                var filter = BuildFilterWithRowKeyBound(cat, dateFrom, dateTo, catContinuation?.LastRowKey, filters);
-
-                var fetched = new List<(string RowKey, OpsEventEntry Item)>();
-                await foreach (var e in _table.QueryAsync<TableEntity>(filter: filter, maxPerPage: pageSize))
+            return PerPartitionFanOutMerge.FetchPageAsync<OpsEventEntry>(
+                AllCategories, pageSize, continuation,
+                async (category, bound, limit, ct) =>
                 {
-                    fetched.Add((e.RowKey, MapToEntry(e)));
-                    if (fetched.Count >= pageSize) break;
-                }
-                return new PerPartitionFanOutMerge.PartitionFetchResult<OpsEventEntry>(cat, fetched);
-            }).ToList();
-
-            var results = await Task.WhenAll(fetchTasks);
-
-            var (items, nextContinuations) = PerPartitionFanOutMerge.MergeAndAdvance(
-                results, continuations, pageSize, e => e.Timestamp);
-
-            // Map only carries active partitions; empty = pagination complete.
-            string? nextRawToken = nextContinuations.Count > 0
-                ? PerPartitionFanOutMerge.EncodeMultiContinuation(nextContinuations)
-                : null;
-            return new RawPage<OpsEventEntry>(items, nextRawToken);
+                    var filter = BuildFilterWithRowKeyBound(category, dateFrom, dateTo, bound, filters);
+                    var read = new List<(string RowKey, OpsEventEntry Item)>();
+                    await foreach (var e in _table.QueryAsync<TableEntity>(filter: filter, maxPerPage: limit, cancellationToken: ct))
+                    {
+                        read.Add((e.RowKey, MapToEntry(e)));
+                        if (read.Count >= limit) break;
+                    }
+                    return read;
+                });
         }
 
         internal static string BuildFilterWithRowKeyBound(
-            string category, DateTime? dateFrom, DateTime? dateTo, string? lastRowKey,
+            string category, DateTime? dateFrom, DateTime? dateTo, PerPartitionFanOutMerge.RowKeyBound? rowKeyBound,
             OpsEventQueryFilters? filters = null)
         {
             var clauses = new List<string>
             {
                 $"PartitionKey eq '{category.Replace("'", "''")}'",
             };
-            if (!string.IsNullOrEmpty(lastRowKey))
-                clauses.Add($"RowKey gt '{lastRowKey!.Replace("'", "''")}'");
+            if (rowKeyBound != null)
+                clauses.Add(rowKeyBound.ToODataClause());
             if (dateFrom.HasValue)
                 clauses.Add(BusinessTimestamp.OpsDateFromClause(ToUtc(dateFrom.Value)));
             if (dateTo.HasValue)

src/Backend/AutopilotMonitor.Functions/Functions/Admin/TenantGroupManagementFunction.cs

@@ -23,8 +23,9 @@ namespace AutopilotMonitor.Functions.Functions.Admin;
 /// repository directly) so the delegated-scope cache is invalidated in lockstep.
 ///
 /// Audit: access-affecting mutations are logged under the <b>affected managed tenant(s)</b>' trail (so the
-/// customer sees "who can read my tenant" / "access removed"). Group create/rename carry no tenant context
-/// and are not customer-visible access changes — they are logged operationally only (no AuditLogs partition).
+/// customer sees "who can read my tenant" / "access removed"); the group appears there under its customer-facing
+/// name (<see cref="DelegationSelfService.CustomerNameOf"/>), never its internal name. Group create/rename/label
+/// carry no access change — they are logged operationally only (no AuditLogs partition).
 /// </summary>
 public class TenantGroupManagementFunction
 {
@@ -100,8 +101,11 @@ public class TenantGroupManagementFunction
     }
 
     /// <summary>
-    /// PATCH /api/global/tenant-groups/{groupId} — rename a group. GlobalAdminOnly.
-    /// Body: { "name": "..." }. A rename has no scope effect.
+    /// PATCH /api/global/tenant-groups/{groupId} — rename a group and/or set the customer label of an operator
+    /// group. GlobalAdminOnly. Body: { "name"?: "...", "customerLabel"?: "..." } (only the fields to change;
+    /// an empty label clears it). Neither has a scope effect; the label only changes how the access is named to
+    /// the member tenants (<see cref="DelegationSelfService.CustomerNameOf"/>), so it is logged operationally
+    /// like a rename, not audited as an access change.
     /// </summary>
     [Function("UpdateTenantGroup")]
     [Authorize]
@@ -114,14 +118,55 @@ public class TenantGroupManagementFunction
         var read = await req.ReadAsync<UpdateTenantGroupRequest>();
         if (read.Error != null) return read.Error;
         var body = read.Value!;
-        if (string.IsNullOrWhiteSpace(body.Name))
-            return await Bad(req, "name is required");
+        var validationError = ValidateUpdateRequest(body);
+        if (validationError != null)
+            return await Bad(req, validationError);
 
-        if (!await _delegatedAdminService.RenameGroupAsync(groupId, body.Name))
-            return await NotFound(req);
-        _logger.LogInformation("Tenant group renamed: {GroupId} -> '{Name}' by {By}", groupId, body.Name, currentUpn);
+        return await UpdateGroupCoreAsync(groupId, body, currentUpn) switch
+        {
+            GroupUpdateOutcome.NotFound => await NotFound(req),
+            GroupUpdateOutcome.LabelOnOwnedGroup => await req.ConflictAsync(
+                "A self-service group has no customer label: its customers see the managing organization's name."),
+            _ => await req.OkAsync(new MessageResponse { Message = "Group updated" }),
+        };
+    }
 
-        return await req.OkAsync(new MessageResponse { Message = "Group updated" });
+    /// <summary>Outcome of <see cref="UpdateGroupCoreAsync"/>.</summary>
+    internal enum GroupUpdateOutcome { Updated, NotFound, LabelOnOwnedGroup }
+
+    /// <summary>Shape check of a PATCH body (400 message, or null when valid): at least one field, no blank name, a bounded label.</summary>
+    internal static string? ValidateUpdateRequest(UpdateTenantGroupRequest body)
+    {
+        if (body.Name == null && body.CustomerLabel == null)
+            return "name or customerLabel is required";
+        if (body.Name != null && string.IsNullOrWhiteSpace(body.Name))
+            return "name must not be empty";
+        if (body.CustomerLabel != null && body.CustomerLabel.Trim().Length > DelegationSelfService.CustomerLabelMaxLength)
+            return $"customerLabel must be at most {DelegationSelfService.CustomerLabelMaxLength} characters";
+        return null;
+    }
+
+    /// <summary>
+    /// Testable core of the group update (body already validated). A customer label is refused on an OWNED
+    /// (self-service) group — its customers see the managing tenant's name, so a label there would be dead data.
+    /// </summary>
+    internal async Task<GroupUpdateOutcome> UpdateGroupCoreAsync(string groupId, UpdateTenantGroupRequest body, string? currentUpn)
+    {
+        if (body.CustomerLabel != null)
+        {
+            var group = await _delegatedAdminService.GetGroupAsync(groupId);
+            if (group == null)
+                return GroupUpdateOutcome.NotFound;
+            if (!string.IsNullOrWhiteSpace(group.OwnerTenantId))
+                return GroupUpdateOutcome.LabelOnOwnedGroup;
+        }
+
+        if (!await _delegatedAdminService.UpdateGroupAsync(groupId, body.Name, body.CustomerLabel))
+            return GroupUpdateOutcome.NotFound;
+
+        _logger.LogInformation("Tenant group updated: {GroupId} name='{Name}' customerLabel='{CustomerLabel}' by {By}",
+            groupId, body.Name, body.CustomerLabel, currentUpn);
+        return GroupUpdateOutcome.Updated;
     }
 
     /// <summary>DELETE /api/global/tenant-groups/{groupId} — delete group + all its assignments. GlobalAdminOnly.</summary>
@@ -173,7 +218,7 @@ public class TenantGroupManagementFunction
             await AuditPerTenantAsync(group.TenantIds, "DELETE", "*", currentUpn,
                 new Dictionary<string, string>
                 {
-                    { "Group", group.Name },
+                    { "Group", DelegationSelfService.CustomerNameOf(group) },
                     { "GroupId", groupId },
                     { "Reason", "group-deleted" },
                     { "AssigneeCount", group.AssigneeCount.ToString() },
@@ -335,7 +380,7 @@ public class TenantGroupManagementFunction
         await AuditPerTenantAsync(group.TenantIds, "CREATE", upn, currentUpn,
             new Dictionary<string, string>
             {
-                { "Group", group.Name },
+                { "Group", DelegationSelfService.CustomerNameOf(group) },
                 { "GroupId", groupId },
                 { "Role", role },
                 { "Reason", "group-assigned" },
@@ -384,7 +429,7 @@ public class TenantGroupManagementFunction
             await AuditPerTenantAsync(group.TenantIds, "DELETE", normalizedUpn, currentUpn,
                 new Dictionary<string, string>
                 {
-                    { "Group", group.Name },
+                    { "Group", DelegationSelfService.CustomerNameOf(group) },
                     { "GroupId", groupId },
                     { "Reason", "group-unassigned" },
                 });

src/Backend/AutopilotMonitor.Functions/Functions/Admin/TenantOffboardFunction.cs

@@ -100,19 +100,22 @@ public class TenantOffboardFunction
             "TENANT OFFBOARD initiated for tenant {TenantId} by {Upn}",
             normalizedTenantId, upn);
 
-        // 2. Capture DomainName from TenantConfiguration so the worker's Completed/Failed
-        //    OpsEvents can render "{domain} ({tenantId})" instead of the bare GUID.
-        string? domainName = null;
+        // 2. Capture what must outlive the TenantConfiguration row the cascade deletes: DomainName
+        //    for the worker's Completed/Failed OpsEvents ("{domain} ({tenantId})"), and TrialConsumed
+        //    for the pointer, which carries it to the returning tenant's first login (D-288).
+        //    Fail-loud: without the read the one-trial-per-tenant rule would silently reset.
+        string? domainName;
+        bool trialConsumed;
         try
         {
             var existingConfig = await _configRepo.GetTenantConfigurationAsync(targetTenantId);
             domainName = existingConfig?.DomainName;
+            trialConsumed = existingConfig?.TrialConsumed ?? false;
         }
         catch (Exception ex)
         {
-            _logger.LogWarning(ex,
-                "Failed to read tenant configuration for {TenantId} before offboard; proceeding without domain name",
-                normalizedTenantId);
+            _logger.LogError(ex, "Failed to read tenant configuration for {TenantId} before offboard", normalizedTenantId);
+            return await Build500Async(req, "Failed to read tenant configuration");
         }
 
         // 2.b — Capture the Preview-Notification-Email before Phase 2.D wipes the
@@ -184,14 +187,14 @@ public class TenantOffboardFunction
         // 5. Upsert ByTenant pointer with ETag-CAS so OffboardCount increments race-safely.
         try
         {
-            await UpsertPointerWithCasAsync(normalizedTenantId, historyRowKey, now);
+            await UpsertPointerWithCasAsync(normalizedTenantId, historyRowKey, now, trialConsumed);
         }
         catch (Exception ex)
         {
             // History row is already committed — operator can find it under
-            // PartitionKey="OffboardingHistory" and retry manually. Graceful degradation
-            // (plan §4.4): pointer is the re-onboarding index; without it PR4's auto-wipe
-            // is no-op but the offboarding itself still proceeds.
+            // PartitionKey="OffboardingHistory". Nothing destructive has started: without the
+            // pointer the returning tenant would get its trial back, so refuse and let the
+            // admin re-click.
             _logger.LogError(ex,
                 "Failed to upsert OffboardingByTenant pointer for {TenantId} after history row inserted",
                 normalizedTenantId);
@@ -722,7 +725,7 @@ public class TenantOffboardFunction
     /// <para>
     /// Drain-deadline math the caller applies:
     /// <c>effectiveDeadline = max(history.EarliestProcessingAt, TombstoneLastUpdated + DrainBarrier)</c>.
-    /// Because the tombstone's <c>LastUpdated</c> is part of the same atomic Save as the
+    /// Because the tombstone's <c>LastUpdated</c> is part of the same atomic write as the
     /// Disabled-flip, it is always present once the auth-gate is committed — even if
     /// the History-row patch later fails on a fast re-click.
     /// </para>
@@ -734,7 +737,7 @@ public class TenantOffboardFunction
     internal enum EnsureDisabledOutcome
     {
         /// <summary>Row was already a Disabled=true tombstone with the offboarding reason —
-        /// no Save was issued. Existing <c>LastUpdated</c> is returned.</summary>
+        /// nothing was written. Existing <c>LastUpdated</c> is returned.</summary>
         AlreadyTombstone,
         /// <summary>We actively wrote the Disabled=true gate (either flipped an existing
         /// row or created a fresh one). <c>LastUpdated</c> is the timestamp of this write.</summary>
@@ -759,70 +762,54 @@ public class TenantOffboardFunction
     /// <para>
     /// Behaviour:
     /// <list type="bullet">
-    ///   <item>Existing TenantConfiguration row → read-modify-write so signup data + custom
-    ///   rate limits etc. are preserved. Idempotent: if already
-    ///   <c>Disabled=true</c> with the offboarding reason, no Save is issued and
+    ///   <item>Existing TenantConfiguration row → conditional write on the fresh row
+    ///   (<see cref="TenantConfigurationService.CreateOrUpdateAsync"/>) so signup data + custom
+    ///   rate limits etc. are preserved and no concurrent write is rewound. Idempotent: if already
+    ///   <c>Disabled=true</c> with the offboarding reason, nothing is written and
     ///   <see cref="EnsureDisabledResult.AlreadyTombstone"/> is returned.</item>
     ///   <item>Missing TenantConfiguration row (edge case: very old tenant / manual delete)
-    ///   → create a fresh <c>CreateDefault</c> row with <c>Disabled=true</c>. Necessary
-    ///   because <c>AuthFunction</c> would otherwise auto-create a <c>Disabled=false</c>
+    ///   → create a fresh <c>CreateDefault</c> row with <c>Disabled=true</c> (conditional insert).
+    ///   Necessary because <c>AuthFunction</c> would otherwise auto-create a <c>Disabled=false</c>
     ///   default on the next <c>/api/auth/me</c> and slip the offboarded tenant back through.</item>
-    ///   <item><c>SaveTenantConfigurationAsync</c> returns <c>false</c> on transient storage
-    ///   failure (the production repo does NOT throw; see Codex-finding TableConfigRepository).
-    ///   We turn that into <see cref="InvalidOperationException"/> so the caller treats it
-    ///   as a hard fail.</item>
+    ///   <item>Every conditional write lost to concurrent changes, or storage failed →
+    ///   throws, so the caller treats it as a hard fail.</item>
     /// </list>
     /// </para>
     /// </summary>
     internal async Task<EnsureDisabledResult> EnsureTenantDisabledAsync(string targetTenantId, string upn)
     {
-        var tenantConfig = await _configRepo.GetTenantConfigurationAsync(targetTenantId);
-
-        if (tenantConfig != null)
+        var update = await _tenantConfigService.CreateOrUpdateAsync(targetTenantId, row =>
         {
-            // Idempotent: if Disabled is already true with our reason, no-op the write so
-            // resume-clicks do not thrash the LastUpdated timestamp on the row.
-            var alreadyDisabled = tenantConfig.Disabled
-                && string.Equals(tenantConfig.DisabledReason, OffboardingDisabledReason, StringComparison.Ordinal);
-            if (alreadyDisabled)
+            if (string.IsNullOrEmpty(row.DomainName))
             {
-                _tenantConfigService.InvalidateCache(targetTenantId);
-                return new EnsureDisabledResult(EnsureDisabledOutcome.AlreadyTombstone, tenantConfig.LastUpdated);
+                // Missing-row edge case: the default is written as the tombstone so AuthFunction's
+                // auto-create-default cannot bring the tenant back enabled.
+                _logger.LogWarning(
+                    "TenantOffboard: tenant configuration for {TenantId} has no domain (missing or never onboarded) — writing the Disabled=true tombstone onto it",
+                    targetTenantId);
             }
-        }
-        else
-        {
-            // Missing-row edge case: create a fresh Disabled=true row so AuthFunction's
-            // auto-create-default cannot bring the tenant back enabled.
-            tenantConfig = TenantConfiguration.CreateDefault(targetTenantId);
-            _logger.LogWarning(
-                "TenantOffboard: no TenantConfiguration row found for {TenantId} — creating fresh Disabled=true tombstone",
-                targetTenantId);
-        }
+            row.Disabled = true;
+            row.DisabledReason = OffboardingDisabledReason;
+            row.DisabledUntil = null;
+            row.UpdatedBy = upn;
+            return true;
+        }, "offboard", "Disabled-gate before cascade");
 
-        tenantConfig.Disabled = true;
-        tenantConfig.DisabledReason = OffboardingDisabledReason;
-        tenantConfig.DisabledUntil = null;
-        tenantConfig.UpdatedBy = upn;
-        // Explicit LastUpdated stamp — we bypass TenantConfigurationService.SaveConfigurationAsync
-        // (which would set this) and write to the repo directly, so we own the bookkeeping.
-        // Critical for the resume-path drain-deadline math: this timestamp is the
-        // authoritative "when did the cache-drain barrier start" anchor.
-        var writtenAt = DateTime.UtcNow;
-        tenantConfig.LastUpdated = writtenAt;
-
-        var saved = await _configRepo.SaveTenantConfigurationAsync(tenantConfig, "offboard", "Disabled-gate before cascade");
-        if (!saved)
+        switch (update.Status)
         {
-            // TableConfigRepository.SaveTenantConfigurationAsync returns false on transient
-            // storage failure instead of throwing. Without throwing here, the caller would
-            // proceed to enqueue without the auth-gate committed — defeating the whole
-            // cache-drain barrier. Surface as a hard failure.
-            throw new InvalidOperationException(
-                $"SaveTenantConfigurationAsync returned false for tenant {targetTenantId}; Disabled-gate NOT committed");
+            case TenantConfigUpdateStatus.OffboardingInProgress:
+                // Idempotent: the tombstone is already in place — nothing written, so LastUpdated
+                // stays the authoritative "when did the cache-drain barrier start" anchor.
+                _tenantConfigService.InvalidateCache(targetTenantId);
+                return new EnsureDisabledResult(EnsureDisabledOutcome.AlreadyTombstone, update.Config!.LastUpdated);
+            case TenantConfigUpdateStatus.Updated:
+                // LastUpdated is stamped by the write itself — the anchor of the drain-deadline math.
+                return new EnsureDisabledResult(EnsureDisabledOutcome.WroteNewTombstone, update.Config!.LastUpdated);
+            default:
+                // Without Disabled=true the cache-drain barrier offers no protection: hard fail.
+                throw new InvalidOperationException(
+                    $"Disabled-gate NOT committed for tenant {targetTenantId} ({update.Status})");
         }
-        _tenantConfigService.InvalidateCache(targetTenantId);
-        return new EnsureDisabledResult(EnsureDisabledOutcome.WroteNewTombstone, writtenAt);
     }
 
     /// <summary>
@@ -869,11 +856,10 @@ public class TenantOffboardFunction
     }
 
     // Plan §4.4 read-modify-write CAS-loop. Bounded retry; throws on exhaustion so the
-    // caller turns it into a 500. The pointer is the O(1) re-onboarding index — failure
-    // here is recoverable (graceful degradation in PR4) but we still surface it so the
-    // operator can investigate.
+    // caller turns it into a 500. The pointer is the O(1) re-onboarding index; TrialConsumed
+    // is sticky across rounds (OR), never reset by an offboarding without a trial.
     internal async Task UpsertPointerWithCasAsync(
-        string normalizedTenantId, string historyRowKey, DateTime now)
+        string normalizedTenantId, string historyRowKey, DateTime now, bool trialConsumed)
     {
         for (var attempt = 1; attempt <= PointerCasMaxAttempts; attempt++)
         {
@@ -890,6 +876,7 @@ public class TenantOffboardFunction
                     LatestStatus = "Initiated",
                     LatestUpdatedAt = now,
                     OffboardCount = 1,
+                    TrialConsumed = trialConsumed,
                 };
 
                 try
@@ -910,6 +897,7 @@ public class TenantOffboardFunction
             existing.LatestStatus = "Initiated";
             existing.LatestUpdatedAt = now;
             existing.OffboardCount = existing.OffboardCount + 1;
+            existing.TrialConsumed |= trialConsumed;
 
             try
             {

src/Backend/AutopilotMonitor.Functions/Functions/Apps/AppsAnalyticsHelper.cs

@@ -30,6 +30,17 @@ namespace AutopilotMonitor.Functions.Functions.Apps
             return Guid.TryParse(raw, out _);
         }
 
+        /// <summary>
+        /// Parses the optional <c>?source=</c> install channel of the per-app endpoints. Absent means
+        /// <see cref="AppInstallSources.Ime"/>, the channel every app link carried before channels
+        /// existed. Returns <c>false</c> for an unknown value (caller should emit a 400).
+        /// </summary>
+        public static bool TryParseSourceQueryParam(string? raw, out string source)
+        {
+            source = string.IsNullOrEmpty(raw) ? AppInstallSources.Ime : raw!.Trim().ToLowerInvariant();
+            return AppInstallSources.IsKnown(source);
+        }
+
         // ── Opt-in pagination ───────────────────────────────────────────────
 
         /// <summary>Upper bound for a single <c>apps/list</c> page when the caller opts into pagination.</summary>
@@ -141,7 +152,9 @@ namespace AutopilotMonitor.Functions.Functions.Apps
             // they are excluded from every per-app group and disclosed via collisionExcluded.
             var collisionExcluded = summaries.Count(s => s.AppIdCollision);
 
-            var apps = summaries.Where(s => !s.AppIdCollision).GroupBy(s => s.AppName).Select(g =>
+            // One group per (channel, name): an Intune app and a RealmJoin package that share a
+            // display name are different installs and stay separate rows.
+            var apps = summaries.Where(s => !s.AppIdCollision).GroupBy(s => (Source: AppInstallSources.Normalize(s.Source), s.AppName)).Select(g =>
             {
                 // PR0 classification (see MetricsMath.IsSkipTerminalState / HasMeasuredDuration):
                 // skips leave the rate + durations; duration stats read measured rows only.
@@ -159,7 +172,8 @@ namespace AutopilotMonitor.Functions.Functions.Apps
 
                 return new AppsListItem
                 {
-                    AppName = g.Key,
+                    AppName = g.Key.AppName,
+                    Source = g.Key.Source,
                     AppType = g.Select(s => s.AppType).FirstOrDefault(t => !string.IsNullOrEmpty(t)) ?? string.Empty,
                     TotalInstalls = total,
                     Succeeded = installed.Count,
@@ -178,6 +192,7 @@ namespace AutopilotMonitor.Functions.Functions.Apps
             .OrderByDescending(a => a.Failed)
             .ThenByDescending(a => a.FailureRate)
             .ThenBy(a => a.AppName, StringComparer.OrdinalIgnoreCase) // deterministic tiebreaker for stable paging cursors
+            .ThenBy(a => a.Source, StringComparer.Ordinal)
             .ToList();
 
             // Legacy mode: caller did not opt into pagination → full array (web UI pages client-side).
@@ -225,6 +240,7 @@ namespace AutopilotMonitor.Functions.Functions.Apps
             List<AppInstallSummary> allSummaries,
             ISessionRepository sessionRepo,
             string appName,
+            string source,
             int days,
             IReadOnlyList<AppVersionRegressionAlert>? versionRegressions = null)
         {
@@ -238,6 +254,7 @@ namespace AutopilotMonitor.Functions.Functions.Apps
 
             var inWindow = allSummaries
                 .Where(s => string.Equals(s.AppName, appName, StringComparison.OrdinalIgnoreCase)
+                            && AppInstallSources.Normalize(s.Source) == source
                             && s.StartedAt >= cutoff)
                 .ToList();
 
@@ -252,6 +269,7 @@ namespace AutopilotMonitor.Functions.Functions.Apps
                 {
                     Success = true,
                     AppName = appName,
+                    Source = source,
                     AppType = string.Empty,
                     WindowDays = days,
                     CollisionExcluded = collisionExcluded,
@@ -413,6 +431,7 @@ namespace AutopilotMonitor.Functions.Functions.Apps
             {
                 Success = true,
                 AppName = appName,
+                Source = source,
                 AppType = appType,
                 WindowDays = days,
                 CollisionExcluded = collisionExcluded,
@@ -449,6 +468,7 @@ namespace AutopilotMonitor.Functions.Functions.Apps
             List<AppInstallSummary> allSummaries,
             ISessionRepository sessionRepo,
             string appName,
+            string source,
             int days,
             string statusFilter,
             string? modelFilter,
@@ -460,6 +480,7 @@ namespace AutopilotMonitor.Functions.Functions.Apps
 
             var summaries = allSummaries
                 .Where(s => string.Equals(s.AppName, appName, StringComparison.OrdinalIgnoreCase)
+                            && AppInstallSources.Normalize(s.Source) == source
                             && s.StartedAt >= cutoff)
                 .ToList();
 

src/Backend/AutopilotMonitor.Functions/Functions/Apps/GetAppAnalyticsFunction.cs

@@ -1,6 +1,7 @@
 using System.Net;
 using AutopilotMonitor.Functions.Helpers;
 using AutopilotMonitor.Shared.DataAccess;
+using AutopilotMonitor.Shared.Models;
 using Microsoft.Azure.Functions.Worker;
 using Microsoft.Azure.Functions.Worker.Http;
 using Microsoft.Extensions.Logging;
@@ -48,14 +49,19 @@ namespace AutopilotMonitor.Functions.Functions.Apps
 
                 var query = req.Query;
                 var days = QueryParams.Int(query["days"], @default: 30, min: 1, max: 365);
+                if (!AppsAnalyticsHelper.TryParseSourceQueryParam(query["source"], out var source))
+                {
+                    return await req.BadRequestAsync($"source must be one of: {string.Join(", ", AppInstallSources.All)}");
+                }
 
                 var summaries = await AppsAnalyticsHelper.LoadSummariesAsync(_metricsRepo, tenantId, days);
                 // Active duration-regression episodes for this app (fail-soft: empty on error).
                 var versionRegressions = (await _notificationTracker.GetAppVersionRegressionsAsync(tenantId))
-                    .Where(a => string.Equals(a.AppName, decodedAppName, StringComparison.OrdinalIgnoreCase))
+                    .Where(a => string.Equals(a.AppName, decodedAppName, StringComparison.OrdinalIgnoreCase)
+                                && AppInstallSources.Normalize(a.Source) == source)
                     .ToList();
                 var body = await AppsAnalyticsHelper.BuildAnalyticsResponseAsync(
-                    summaries, _sessionRepo, decodedAppName, days, versionRegressions);
+                    summaries, _sessionRepo, decodedAppName, source, days, versionRegressions);
 
                 var response = req.CreateResponse(HttpStatusCode.OK);
                 await response.WriteAsJsonAsync(body);

src/Backend/AutopilotMonitor.Functions/Functions/Apps/GetAppSessionsFunction.cs

@@ -1,6 +1,7 @@
 using System.Net;
 using AutopilotMonitor.Functions.Helpers;
 using AutopilotMonitor.Shared.DataAccess;
+using AutopilotMonitor.Shared.Models;
 using Microsoft.Azure.Functions.Worker;
 using Microsoft.Azure.Functions.Worker.Http;
 using Microsoft.Extensions.Logging;
@@ -47,6 +48,10 @@ namespace AutopilotMonitor.Functions.Functions.Apps
 
                 var query = req.Query;
                 var days = QueryParams.Int(query["days"], @default: 30, min: 1, max: 365);
+                if (!AppsAnalyticsHelper.TryParseSourceQueryParam(query["source"], out var source))
+                {
+                    return await req.BadRequestAsync($"source must be one of: {string.Join(", ", AppInstallSources.All)}");
+                }
 
                 var statusFilter = (query["status"] ?? "all").Trim().ToLowerInvariant();
                 var modelFilter = query["model"];
@@ -58,7 +63,7 @@ namespace AutopilotMonitor.Functions.Functions.Apps
 
                 var summaries = await AppsAnalyticsHelper.LoadSummariesAsync(_metricsRepo, tenantId, days);
                 var body = await AppsAnalyticsHelper.BuildSessionsResponseAsync(
-                    summaries, _sessionRepo, decodedAppName, days,
+                    summaries, _sessionRepo, decodedAppName, source, days,
                     statusFilter, modelFilter, versionFilter, offset, limit);
 
                 var response = req.CreateResponse(HttpStatusCode.OK);

src/Backend/AutopilotMonitor.Functions/Functions/Apps/GetGlobalAppAnalyticsFunction.cs

@@ -53,6 +53,10 @@ namespace AutopilotMonitor.Functions.Functions.Apps
                     return await req.BadRequestAsync("tenantId must be a valid GUID");
                 }
                 var days = QueryParams.Int(query["days"], @default: 30, min: 1, max: 365);
+                if (!AppsAnalyticsHelper.TryParseSourceQueryParam(query["source"], out var source))
+                {
+                    return await req.BadRequestAsync($"source must be one of: {string.Join(", ", AppInstallSources.All)}");
+                }
 
                 _logger.LogInformation(
                     "Global apps/{App}/analytics requested (user: {User}, tenantId: {TenantId}, days: {Days})",
@@ -64,10 +68,11 @@ namespace AutopilotMonitor.Functions.Functions.Apps
                 var versionRegressions = string.IsNullOrEmpty(scopedTenantId)
                     ? new List<AppVersionRegressionAlert>()
                     : (await _notificationTracker.GetAppVersionRegressionsAsync(scopedTenantId!))
-                        .Where(a => string.Equals(a.AppName, decodedAppName, StringComparison.OrdinalIgnoreCase))
+                        .Where(a => string.Equals(a.AppName, decodedAppName, StringComparison.OrdinalIgnoreCase)
+                                    && AppInstallSources.Normalize(a.Source) == source)
                         .ToList();
                 var body = await AppsAnalyticsHelper.BuildAnalyticsResponseAsync(
-                    summaries, _sessionRepo, decodedAppName, days, versionRegressions);
+                    summaries, _sessionRepo, decodedAppName, source, days, versionRegressions);
 
                 var response = req.CreateResponse(HttpStatusCode.OK);
                 await response.WriteAsJsonAsync(body);

src/Backend/AutopilotMonitor.Functions/Functions/Apps/GetGlobalAppSessionsFunction.cs

@@ -1,6 +1,7 @@
 using System.Net;
 using AutopilotMonitor.Functions.Helpers;
 using AutopilotMonitor.Shared.DataAccess;
+using AutopilotMonitor.Shared.Models;
 using Microsoft.Azure.Functions.Worker;
 using Microsoft.Azure.Functions.Worker.Http;
 using Microsoft.Extensions.Logging;
@@ -52,6 +53,10 @@ namespace AutopilotMonitor.Functions.Functions.Apps
                     return await req.BadRequestAsync("tenantId must be a valid GUID");
                 }
                 var days = QueryParams.Int(query["days"], @default: 30, min: 1, max: 365);
+                if (!AppsAnalyticsHelper.TryParseSourceQueryParam(query["source"], out var source))
+                {
+                    return await req.BadRequestAsync($"source must be one of: {string.Join(", ", AppInstallSources.All)}");
+                }
 
                 var statusFilter = (query["status"] ?? "all").Trim().ToLowerInvariant();
                 var modelFilter = query["model"];
@@ -67,7 +72,7 @@ namespace AutopilotMonitor.Functions.Functions.Apps
 
                 var summaries = await AppsAnalyticsHelper.LoadSummariesAsync(_metricsRepo, scopedTenantId, days);
                 var body = await AppsAnalyticsHelper.BuildSessionsResponseAsync(
-                    summaries, _sessionRepo, decodedAppName, days,
+                    summaries, _sessionRepo, decodedAppName, source, days,
                     statusFilter, modelFilter, versionFilter, offset, limit);
 
                 var response = req.CreateResponse(HttpStatusCode.OK);

src/Backend/AutopilotMonitor.Functions/Functions/Config/PatchTenantConfigurationFieldsFunction.cs

@@ -17,15 +17,25 @@ namespace AutopilotMonitor.Functions.Functions.Config
     /// <summary>
     /// Transactional field-level patch of a tenant's configuration (TenantAdminOrGA: a tenant
     /// admin patches their own row on the stricter TenantAdmin caller tier — GA-only fields are
-    /// an explicit 400 there). Unlike the full-model PUT, this takes ONLY the fields to change,
-    /// and the service verifies after the conditional write that exactly those fields changed —
-    /// rolling back automatically on drift. Every write is preceded by a fail-closed snapshot
-    /// into ConfigurationBackups (revertible via POST config/{tenantId}/revert, GA-only).
+    /// an explicit 400 there). The one client write path of the tenant configuration (the
+    /// full-model PUT is gone, D-290): it takes ONLY the fields to change, and the service
+    /// verifies after the conditional write that exactly those fields changed — rolling back
+    /// automatically on drift. Every write is preceded by a fail-closed snapshot into
+    /// ConfigurationBackups (revertible via POST config/{tenantId}/revert, GA-only).
     /// </summary>
     public class PatchTenantConfigurationFieldsFunction
     {
         internal const int MaxBodyBytes = 65_536;
 
+        /// <summary>
+        /// Write-path label for the session-detail Collect Logs quick-config dialog. The dialog
+        /// sends <c>?intent=collect-logs</c> so the backup snapshot and the ops event can name the
+        /// path (<see cref="OpsEventService"/> raises its own event type for it). Allow-listed,
+        /// never echoed from arbitrary input.
+        /// </summary>
+        internal const string CollectLogsIntent = "collect-logs";
+        internal const string CollectLogsSource = "portal-collect-logs";
+
         private readonly ILogger<PatchTenantConfigurationFieldsFunction> _logger;
         private readonly TenantConfigPatchService _patchService;
 
@@ -86,12 +96,19 @@ namespace AutopilotMonitor.Functions.Functions.Config
             }
         }
 
-        /// <summary>Tags the backup Source with the write path (MCP tools stamp X-Client-Source: mcp).</summary>
+        /// <summary>
+        /// Tags the backup Source with the write path: the Collect Logs intent, else MCP (tools stamp
+        /// X-Client-Source: mcp) or the plain API.
+        /// </summary>
         internal static string ResolveSource(HttpRequestData req, string operation)
-            => req.Headers.TryGetValues("X-Client-Source", out var values)
-               && string.Equals(values.FirstOrDefault(), "mcp", StringComparison.OrdinalIgnoreCase)
+        {
+            if (operation == "patch" && string.Equals(req.Query["intent"], CollectLogsIntent, StringComparison.OrdinalIgnoreCase))
+                return CollectLogsSource;
+            return req.Headers.TryGetValues("X-Client-Source", out var values)
+                   && string.Equals(values.FirstOrDefault(), "mcp", StringComparison.OrdinalIgnoreCase)
                 ? $"mcp-{operation}"
                 : $"api-{operation}";
+        }
 
         internal static async Task<HttpResponseData> WriteOutcome(HttpRequestData req, PatchOutcome outcome)
         {