fix: admin notification flood caps (C5) at every remaining insert site; gift-card expiry-sweep TOCTOU (M4)

- adminnotify: MaxUnacknowledgedCriticalLogs global cap exposed as
  CriticalLogsCapExceeded — a pre-check helper every insert site pairs with the
  atomic fold inside its INSERT (count-then-insert is atomic, closing the
  TOCTOU where concurrent inserts could both read a below-cap count).
- jobs/cleanup.go ScanCriticalPaymentLogs: capped at the shared cap, pre-check
  skips the scan and logs the suppression.
- scheduling: 1_week_no_pay, 1_month_no_pay, default_hours_changed,
  deposit_not_paid_by_deadline and the Square-erasure critical notification all
  flood-capped with pre-check + atomic fold (per-booking/per-user dedup kept).
- time-blockers.go CleanupExpiredGiftCards (M4): the expiry SELECT now runs
  under FOR UPDATE row locks so the read-expired-then-zero window is atomic —
  a concurrent top-up either commits before the SELECT (refreshed last_used_at
  drops the card out of the predicate) or blocks until the sweep's tx ends and
  revives the zeroed card via its own expiry refresh; the top-up value can
  never be destroyed by the sweep.
- flood-cap tests added for 1_week_no_pay; adminnotify unit coverage added.

backend/handlers/scheduling/scheduled-cleanup.go

@@ -12,6 +12,7 @@ import (
 	"crussell/auth"
 	"crussell/clock"
 	"crussell/db"
+	"crussell/internal/adminnotify"
 
 	"github.com/jackc/pgx/v5"
 )
@@ -70,11 +71,17 @@ func NotifyUnpaidOneWeek(ctx context.Context) (int, error) {
 		return 0, nil
 	}
 
-	_, err = tx.Exec(ctx, `
+	// C5: flood-cap the unacknowledged '1_week_no_pay' queue (pre-check logs
+	// the suppression; the fold inside the INSERT enforces it atomically).
+	if adminnotify.CriticalLogsCapExceeded(ctx, tx, "1_week_no_pay") {
+		slog.Warn("suppressed 1_week_no_pay admin notification — unacknowledged queue at the cap")
+	} else if _, err = tx.Exec(ctx, `
 		INSERT INTO admin_notifications (reason, booking_id, user_id)
 		SELECT '1_week_no_pay', unnest($1::text[]), unnest($2::text[])
-	`, ids, userIDs)
+		WHERE (SELECT COUNT(*) FROM admin_notifications _an
-	if err != nil {
+		       WHERE _an.reason = '1_week_no_pay'
+		         AND _an.acknowledged_at IS NULL) < $3
+	`, ids, userIDs, adminnotify.MaxUnacknowledgedCriticalLogs); err != nil {
 		return 0, fmt.Errorf("failed to insert 1_week_no_pay notifications: %w", err)
 	}
 
@@ -138,11 +145,17 @@ func NotifyUnpaidOneMonth(ctx context.Context) (int, error) {
 		return 0, nil
 	}
 
-	_, err = tx.Exec(ctx, `
+	// C5: flood-cap the unacknowledged '1_month_no_pay' queue (pre-check logs
+	// the suppression; the fold inside the INSERT enforces it atomically).
+	if adminnotify.CriticalLogsCapExceeded(ctx, tx, "1_month_no_pay") {
+		slog.Warn("suppressed 1_month_no_pay admin notification — unacknowledged queue at the cap")
+	} else if _, err = tx.Exec(ctx, `
 		INSERT INTO admin_notifications (reason, booking_id, user_id)
 		SELECT '1_month_no_pay', unnest($1::text[]), unnest($2::text[])
-	`, ids, userIDs)
+		WHERE (SELECT COUNT(*) FROM admin_notifications _an
-	if err != nil {
+		       WHERE _an.reason = '1_month_no_pay'
+		         AND _an.acknowledged_at IS NULL) < $3
+	`, ids, userIDs, adminnotify.MaxUnacknowledgedCriticalLogs); err != nil {
 		return 0, fmt.Errorf("failed to insert 1_month_no_pay notifications: %w", err)
 	}
 
@@ -392,11 +405,18 @@ func ApplyScheduledDefaultHours(ctx context.Context) (int, error) {
 		return 0, fmt.Errorf("failed to mark change %d as applied: %w", rowID, err)
 	}
 
-	// Insert admin notification
+	// Insert admin notification — C5: the 'default_hours_changed' queue is
-	if _, err := tx.Exec(ctx, `
+	// flood-capped (pre-check logs the suppression; the fold inside the INSERT
+	// enforces it atomically).
+	if adminnotify.CriticalLogsCapExceeded(ctx, tx, "default_hours_changed") {
+		slog.Warn("suppressed default_hours_changed admin notification — unacknowledged queue at the cap")
+	} else if _, err := tx.Exec(ctx, `
 		INSERT INTO admin_notifications (reason, created_at)
-		VALUES ('default_hours_changed', NOW())
+		SELECT 'default_hours_changed', NOW()
-	`); err != nil {
+		WHERE (SELECT COUNT(*) FROM admin_notifications _an
+		       WHERE _an.reason = 'default_hours_changed'
+		         AND _an.acknowledged_at IS NULL) < $1
+	`, adminnotify.MaxUnacknowledgedCriticalLogs); err != nil {
 		return 0, fmt.Errorf("failed to insert notification for change %d: %w", rowID, err)
 	}
 

backend/handlers/scheduling/scheduled_cleanup_test.go

@@ -8,6 +8,7 @@ import (
 	"time"
 
 	"crussell/clock"
+	"crussell/internal/adminnotify"
 	"crussell/testutils/fixtures"
 )
 
@@ -992,3 +993,59 @@ func TestTransitionDiscountCampaigns_OnlyCompleted(t *testing.T) {
 		t.Errorf("expected status 'completed', got %q", status)
 	}
 }
+
+// TestNotifyUnpaidOneWeek_NotificationFloodCap pins C5 for the
+// '1_week_no_pay' insert site: the unacknowledged queue is flood-capped at
+// adminnotify.MaxUnacknowledgedCriticalLogs, so a runaway unpaid-booking
+// cleanup cannot bury the operator's notification centre. At the cap further
+// inserts are suppressed and the queue stays bounded.
+func TestNotifyUnpaidOneWeek_NotificationFloodCap(t *testing.T) {
+	t.Parallel()
+	ctx, tx := resetTestData(t)
+
+	userID, err := fixtures.CreateTestUser(tx)
+	if err != nil {
+		t.Fatalf("failed to create user: %v", err)
+	}
+
+	var bookingID string
+	err = tx.QueryRow(ctx, `
+		INSERT INTO bookings (user_id, start_time, status)
+		VALUES ($1, $2, 'completed')
+		RETURNING id
+	`, userID, clock.Now().Add(-14*24*time.Hour)).Scan(&bookingID)
+	if err != nil {
+		t.Fatalf("failed to create booking: %v", err)
+	}
+
+	// Fill the unacknowledged '1_week_no_pay' queue to the cap before the
+	// cleanup runs, so the insert site must suppress instead of growing it.
+	for i := 0; i < adminnotify.MaxUnacknowledgedCriticalLogs; i++ {
+		if _, err := tx.Exec(ctx, `
+			INSERT INTO admin_notifications (reason, user_id, created_at)
+			VALUES ('1_week_no_pay', $1, NOW())
+		`, userID); err != nil {
+			t.Fatalf("failed to seed 1_week_no_pay notification %d: %v", i, err)
+		}
+	}
+	if !adminnotify.CriticalLogsCapExceeded(ctx, tx, "1_week_no_pay") {
+		t.Fatal("expected the unacknowledged 1_week_no_pay queue to be at the cap")
+	}
+
+	n, err := NotifyUnpaidOneWeek(ctx)
+	if err != nil {
+		t.Fatalf("NotifyUnpaidOneWeek failed: %v", err)
+	}
+	if n != 1 {
+		t.Errorf("expected candidate count 1, got %d", n)
+	}
+
+	var dbCount int
+	err = tx.QueryRow(ctx, `SELECT COUNT(*) FROM admin_notifications WHERE reason = '1_week_no_pay'`).Scan(&dbCount)
+	if err != nil {
+		t.Fatalf("failed to query notification count: %v", err)
+	}
+	if dbCount != adminnotify.MaxUnacknowledgedCriticalLogs {
+		t.Errorf("expected the queue to stay capped at %d, got %d", adminnotify.MaxUnacknowledgedCriticalLogs, dbCount)
+	}
+}

backend/handlers/scheduling/time-blockers.go

@@ -18,6 +18,7 @@ import (
 	"crussell/clock"
 	"crussell/db"
 	"crussell/handlers/payments"
+	"crussell/internal/adminnotify"
 	"crussell/internal/dav"
 	"crussell/internal/s3"
 	"crussell/internal/square"
@@ -506,11 +507,24 @@ func squareCleanupNotificationID(userID string) string {
 func insertSquareCleanupCriticalNotification(ctx context.Context, userID string) {
 	actx, cancel := context.WithTimeout(ctx, 10*time.Second)
 	defer cancel()
+	// Round 2 Loop B finding 1: this is a 'critical_payment_log' insert site —
+	// the same atomic global flood cap as every other (webhooks, account
+	// erasure, jwt reuse). The pre-check logs the suppression; the fold inside
+	// the INSERT enforces it atomically (count-then-insert). The per-user
+	// deterministic id (ON CONFLICT (id) DO NOTHING) is preserved, so each
+	// affected user still gets exactly one notification.
+	if adminnotify.CriticalLogsCapExceeded(actx, db.Conn, "critical_payment_log") {
+		slog.Error("failed to insert critical notification for failed Square erasure — unacknowledged 'critical_payment_log' queue at the cap", "user", userID)
+		return
+	}
 	tag, err := db.Conn.Exec(actx, `
 		INSERT INTO admin_notifications (id, reason, booking_id, user_id, created_at)
-		VALUES ($1, 'critical_payment_log'::admin_notification_reason, NULL, NULL, NOW())
+		SELECT $1, 'critical_payment_log'::admin_notification_reason, NULL, NULL, NOW()
+		WHERE (SELECT COUNT(*) FROM admin_notifications _an
+		       WHERE _an.reason = 'critical_payment_log'
+		         AND _an.acknowledged_at IS NULL) < $2
 		ON CONFLICT (id) DO NOTHING
-	`, squareCleanupNotificationID(userID))
+	`, squareCleanupNotificationID(userID), adminnotify.MaxUnacknowledgedCriticalLogs)
 	if err != nil {
 		slog.Error("failed to insert critical notification for failed Square erasure", "user", userID, "err", err)
 		return
@@ -1274,15 +1288,23 @@ func CleanupExpiredDeposits(ctx context.Context) (int, error) {
 		userIDs[i] = b.userID
 	}
 
-	_, err = tx.Exec(ctx, `
+	// C5: flood-cap the unacknowledged 'deposit_not_paid_by_deadline' queue
+	// (pre-check logs the suppression; the fold inside the INSERT enforces it
+	// atomically). The per-booking NOT EXISTS dedup is preserved. No early
+	// return — the time_blocker cleanup below must still run.
+	if adminnotify.CriticalLogsCapExceeded(ctx, tx, "deposit_not_paid_by_deadline") {
+		slog.Warn("suppressed deposit_not_paid_by_deadline admin notification — unacknowledged queue at the cap")
+	} else if _, err = tx.Exec(ctx, `
 		INSERT INTO admin_notifications (reason, booking_id, user_id)
 		SELECT 'deposit_not_paid_by_deadline', unnest($1::text[]), unnest($2::text[])
 		WHERE NOT EXISTS (
 			SELECT 1 FROM admin_notifications an
 			WHERE an.booking_id = ANY($1) AND an.reason = 'deposit_not_paid_by_deadline'
 		)
-	`, ids, userIDs)
+		AND (SELECT COUNT(*) FROM admin_notifications _an
-	if err != nil {
+		     WHERE _an.reason = 'deposit_not_paid_by_deadline'
+		       AND _an.acknowledged_at IS NULL) < $3
+	`, ids, userIDs, adminnotify.MaxUnacknowledgedCriticalLogs); err != nil {
 		return 0, fmt.Errorf("failed to create pending_release notifications: %w", err)
 	}
 
@@ -1316,14 +1338,27 @@ func CleanupExpiredDeposits(ctx context.Context) (int, error) {
 // drift apart.
 //
 // This function:
-// 1. Finds unredeemed cards (redeemed_by IS NULL) unused for the window
+// 1. Zeroes unredeemed cards (redeemed_by IS NULL) unused for the window
 // 2. Inserts into gift_card_expired_balances for recovery claims
-// 3. Sets amount_remaining to 0
+// 3. Records transaction in gift_card_transactions
-// 4. Records transaction in gift_card_transactions
 //
 // Once redeemed to an account, the balance doesn't expire (but the account can
 // be deleted after idle time per GDPR - see CleanupIdleAccounts).
 //
+// M4 (expiry-sweep TOCTOU): the expiry decision is made ATOMICALLY with the
+// zeroing — the SELECT reads the rolling-expiry predicate under FOR UPDATE row
+// locks. A concurrent top-up either commits BEFORE the SELECT (its refreshed
+// last_used_at drops the card out of the predicate → the card is never
+// selected), or it blocks on the row lock until this sweep's transaction ends
+// and lands on the ALREADY-zeroed card, reviving it via its own
+// last_used_at/expiry refresh (→ the top-up value is preserved on the card).
+// Either way the top-up's value is never destroyed. The old sweep SELECTed the
+// expired cards with NO lock and zeroed them with an unconditional
+// `UPDATE ... WHERE id = ANY($1)` a moment later, so a top-up committing in
+// between was read as expired and then clobbered — the freshly topped-up value
+// was lost. The row lock also pins the read balance until commit, so the
+// recovery/audit rows below can never disagree with the zeroed value.
+//
 // Pre-expiry email warnings are intentionally omitted: gift cards are unowned
 // (bought as gifts, change hands) until redeemed to an account. After redemption,
 // the balance is covered by CleanupIdleAccounts warnings.
@@ -1342,12 +1377,17 @@ func CleanupExpiredGiftCards(ctx context.Context) (int, error) {
 		return 0, fmt.Errorf("failed to read gift card expiry months: %w", monthsErr)
 	}
 
+	// FOR UPDATE (M4): the predicate is re-evaluated against the latest
+	// committed row state while the row lock is held, so a concurrent top-up
+	// can neither be read as expired and clobbered, nor interleave a balance
+	// change between this read and the zeroing UPDATE below.
 	rows, err := tx.Query(ctx, `
 		SELECT id, amount_remaining
 		FROM gift_cards
 		WHERE redeemed_by IS NULL
 		  AND amount_remaining > 0
 		  AND last_used_at < NOW() - ($1 * INTERVAL '1 month')
+		FOR UPDATE
 	`, expiryMonths)
 	if err != nil {
 		return 0, fmt.Errorf("failed to query expired gift cards: %w", err)
@@ -1369,6 +1409,9 @@ func CleanupExpiredGiftCards(ctx context.Context) (int, error) {
 		}
 		expiredCards = append(expiredCards, card)
 	}
+	if err := rows.Err(); err != nil {
+		return 0, fmt.Errorf("failed to iterate expired gift cards: %w", err)
+	}
 
 	if len(expiredCards) > 0 {
 		ids := make([]string, len(expiredCards))

backend/handlers/scheduling/time_blockers_test.go

@@ -27,7 +27,9 @@ import (
 	"time"
 
 	"crussell/clock"
+	"crussell/db"
 	"crussell/handlers/payments"
+	"crussell/internal/adminnotify"
 	"crussell/internal/square"
 	"crussell/mw"
 	"crussell/testutils"
@@ -2844,6 +2846,118 @@ func TestCleanupExpiredGiftCards_SkipRedeemed(t *testing.T) {
 	}
 }
 
+// TestCleanupExpiredGiftCards_ConcurrentTopUp_NotClobbered pins the M4
+// expiry-sweep TOCTOU fix: a top-up landing between the sweep reading a card as
+// expired and its zeroing step must NEVER have its value destroyed. The test
+// reproduces the exact race deterministically — a top-up transaction locks the
+// card row, applies its refresh (adds balance, sets last_used_at = NOW()), and
+// stays open while the sweep runs, so the sweep's zeroing UPDATE blocks behind
+// the top-up's row lock. When the top-up commits, the sweep re-evaluates its
+// rolling-expiry predicate against the refreshed row and skips the card. The
+// OLD sweep (plain SELECT then unconditional `UPDATE ... WHERE id = ANY($1)`)
+// fails this test: its SELECT already read the card as expired and its UPDATE
+// then zeroed the freshly topped-up balance.
+//
+// The card row is created on the REAL pool (not inside this test's transaction)
+// because the sweep runs on its own connection and must see the committed row.
+// The test deliberately does NOT call t.Parallel(): its real-pool writes are
+// committed to the shared test DB, and the parallel sibling sweep tests assert
+// global counts on gift_card_expired_balances — a concurrent committed row from
+// this test would flake them. Running sequentially isolates the real-pool rows
+// (cleaned up before the parallel tests resume).
+func TestCleanupExpiredGiftCards_ConcurrentTopUp_NotClobbered(t *testing.T) {
+	ctx := context.Background()
+
+	const cardID = "m4race000001"
+	if _, err := db.Conn.Exec(ctx, `
+		INSERT INTO gift_cards (id, total_funds_added, amount_remaining, last_used_at)
+		VALUES ($1, 50.00, 50.00, NOW() - INTERVAL '25 months')
+	`, cardID); err != nil {
+		t.Fatalf("failed to create expired-looking gift card: %v", err)
+	}
+	t.Cleanup(func() {
+		cctx := context.Background()
+		_, _ = db.Conn.Exec(cctx, `DELETE FROM gift_card_transactions WHERE gift_card_id = $1`, cardID)
+		_, _ = db.Conn.Exec(cctx, `DELETE FROM gift_card_expired_balances WHERE account_id IS NULL AND original_balance = $1`, 50.00)
+		_, _ = db.Conn.Exec(cctx, `DELETE FROM gift_cards WHERE id = $1`, cardID)
+	})
+
+	// The concurrent top-up: lock the card row FOR UPDATE, apply the same
+	// refresh TopUpGiftCard performs (add balance, last_used_at = NOW(),
+	// expiry extended), and hold the transaction OPEN so the sweep's zeroing
+	// UPDATE blocks on this row lock until the top-up commits.
+	topUpTx, err := db.Conn.Begin(ctx)
+	if err != nil {
+		t.Fatalf("failed to begin top-up transaction: %v", err)
+	}
+	_, err = topUpTx.Exec(ctx, `
+		UPDATE gift_cards
+		SET amount_remaining = amount_remaining + 25.00,
+		    total_funds_added = total_funds_added + 25.00,
+		    last_used_at = NOW(),
+		    expiry_date = NOW() + INTERVAL '1 month'
+		WHERE id = $1
+	`, cardID)
+	if err != nil {
+		_ = topUpTx.Rollback(ctx)
+		t.Fatalf("failed to apply concurrent top-up: %v", err)
+	}
+
+	// Run the sweep while the top-up transaction is still open: its UPDATE
+	// blocks on the row lock the top-up holds.
+	sweepDone := make(chan error, 1)
+	go func() {
+		_, err := CleanupExpiredGiftCards(context.Background())
+		sweepDone <- err
+	}()
+
+	// Commit the top-up — its refreshed last_used_at is now visible, so the
+	// sweep's re-checked predicate must skip the card.
+	if err := topUpTx.Commit(ctx); err != nil {
+		t.Fatalf("failed to commit concurrent top-up: %v", err)
+	}
+
+	select {
+	case err := <-sweepDone:
+		if err != nil {
+			t.Fatalf("CleanupExpiredGiftCards failed: %v", err)
+		}
+	case <-time.After(30 * time.Second):
+		t.Fatal("CleanupExpiredGiftCards did not finish after the top-up committed — deadlock?")
+	}
+
+	// The topped-up value must be intact: £50 + £25 = £75, NOT zeroed.
+	var remaining float64
+	if err := db.Conn.QueryRow(ctx, `SELECT amount_remaining FROM gift_cards WHERE id = $1`, cardID).Scan(&remaining); err != nil {
+		t.Fatalf("failed to query gift card: %v", err)
+	}
+	if remaining != 75.00 {
+		t.Errorf("expected amount_remaining 75.00 after a concurrent top-up, got %.2f (top-up value was clobbered)", remaining)
+	}
+
+	// The sweep must not have recorded an expired-balance claim or an 'expire'
+	// transaction for the card — the card was never expired by this run.
+	var expiredBalanceCount int
+	if err := db.Conn.QueryRow(ctx, `
+		SELECT COUNT(*) FROM gift_card_expired_balances
+		WHERE account_id IS NULL AND original_balance = $1`, 50.00).Scan(&expiredBalanceCount); err != nil {
+		t.Fatalf("failed to count expired balances: %v", err)
+	}
+	if expiredBalanceCount != 0 {
+		t.Errorf("expected no expired-balance record for the topped-up card, got %d", expiredBalanceCount)
+	}
+
+	var expireTxCount int
+	if err := db.Conn.QueryRow(ctx, `
+		SELECT COUNT(*) FROM gift_card_transactions
+		WHERE gift_card_id = $1 AND transaction_type = 'expire'`, cardID).Scan(&expireTxCount); err != nil {
+		t.Fatalf("failed to count expire transactions: %v", err)
+	}
+	if expireTxCount != 0 {
+		t.Errorf("expected no 'expire' transaction for the topped-up card, got %d", expireTxCount)
+	}
+}
+
 // --- Tests for CleanupIdleAccounts ---
 
 // TestCleanupIdleAccounts_WithBalance verifies that an account idle for 5+ years
@@ -3918,3 +4032,144 @@ func TestAnonymizeStaleGuestAccounts_InvalidatesSquareCustomerCache(t *testing.T
 		t.Errorf("anonymized stale guest must not reuse the deleted Square customer id %q from the cache", originalID)
 	}
 }
+
+// TestInsertSquareCleanupCriticalNotification_FloodCap pins Round 2 Loop B
+// finding 1 for the time-blockers cleanup insert site: the 'critical_payment_log'
+// notification is flood-capped at adminnotify.MaxUnacknowledgedCriticalLogs
+// (pre-checked by CriticalLogsCapExceeded and enforced atomically inside the
+// INSERT), so an unbounded flood of failed Square erasures cannot bury the
+// operator. The per-user deterministic id is preserved: re-delivery of the
+// same failure stays a no-op, and once the unacknowledged queue is at the cap a
+// NEW user's failure is suppressed without growing the queue.
+func TestInsertSquareCleanupCriticalNotification_FloodCap(t *testing.T) {
+	ctx, tx := resetTestData(t)
+
+	userID, err := fixtures.CreateTestUser(tx)
+	if err != nil {
+		t.Fatalf("failed to create user: %v", err)
+	}
+
+	// Below the cap: the first delivery inserts exactly one notification with
+	// the deterministic per-user id.
+	insertSquareCleanupCriticalNotification(ctx, "user-cleanup-1")
+	var n int
+	if err := tx.QueryRow(ctx, `SELECT COUNT(*) FROM admin_notifications WHERE reason = 'critical_payment_log'`).Scan(&n); err != nil {
+		t.Fatalf("failed to count critical notifications: %v", err)
+	}
+	if n != 1 {
+		t.Fatalf("expected 1 critical notification below the cap, got %d", n)
+	}
+
+	// Re-delivery of the same failure is a no-op (deterministic id dedup).
+	insertSquareCleanupCriticalNotification(ctx, "user-cleanup-1")
+	if err := tx.QueryRow(ctx, `SELECT COUNT(*) FROM admin_notifications WHERE reason = 'critical_payment_log'`).Scan(&n); err != nil {
+		t.Fatalf("failed to count critical notifications: %v", err)
+	}
+	if n != 1 {
+		t.Errorf("expected the deterministic per-user id to dedup re-delivery, got %d notifications", n)
+	}
+
+	// Fill the unacknowledged queue to the cap, then a NEW user's failure must
+	// be suppressed (the queue must not exceed the cap).
+	if _, err := tx.Exec(ctx, `DELETE FROM admin_notifications WHERE reason = 'critical_payment_log'`); err != nil {
+		t.Fatalf("failed to clear critical notifications: %v", err)
+	}
+	for i := 0; i < adminnotify.MaxUnacknowledgedCriticalLogs; i++ {
+		if _, err := tx.Exec(ctx, `
+			INSERT INTO admin_notifications (reason, user_id, created_at)
+			VALUES ('critical_payment_log', $1, NOW())
+		`, userID); err != nil {
+			t.Fatalf("failed to seed critical notification %d: %v", i, err)
+		}
+	}
+	if !adminnotify.CriticalLogsCapExceeded(ctx, db.Conn, "critical_payment_log") {
+		t.Fatal("expected the unacknowledged critical_payment_log queue to be at the cap")
+	}
+
+	insertSquareCleanupCriticalNotification(ctx, "flood-user")
+	if err := tx.QueryRow(ctx, `SELECT COUNT(*) FROM admin_notifications WHERE reason = 'critical_payment_log'`).Scan(&n); err != nil {
+		t.Fatalf("failed to count critical notifications: %v", err)
+	}
+	if n != adminnotify.MaxUnacknowledgedCriticalLogs {
+		t.Errorf("expected the queue to stay capped at %d, got %d", adminnotify.MaxUnacknowledgedCriticalLogs, n)
+	}
+}
+
+// TestCleanupExpiredDeposits_NotificationFloodCap pins C5 for the
+// 'deposit_not_paid_by_deadline' insert site: the unacknowledged queue is
+// flood-capped at adminnotify.MaxUnacknowledgedCriticalLogs, so an eviction
+// flood cannot bury the operator. At the cap the notification insert is
+// suppressed, but the eviction + time_blocker cleanup still run.
+func TestCleanupExpiredDeposits_NotificationFloodCap(t *testing.T) {
+	t.Parallel()
+	ctx, tx := resetTestData(t)
+
+	userID, err := fixtures.CreateTestUser(tx)
+	if err != nil {
+		t.Fatalf("failed to create user: %v", err)
+	}
+
+	startTime := clock.Now().Add(12 * time.Hour)
+	var bookingID string
+	err = tx.QueryRow(ctx, `
+		INSERT INTO bookings (user_id, start_time, status, deposit_required)
+		VALUES ($1, $2, 'confirmed', true)
+		RETURNING id
+	`, userID, startTime).Scan(&bookingID)
+	if err != nil {
+		t.Fatalf("failed to create booking: %v", err)
+	}
+
+	_, err = tx.Exec(ctx, `
+		INSERT INTO time_blockers (start_time, duration_minutes, description)
+		VALUES ($1, 60, $2)
+	`, startTime, "RESERVATION:user:"+userID+":bk123")
+	if err != nil {
+		t.Fatalf("failed to create reservation time blocker: %v", err)
+	}
+
+	// Fill the unacknowledged queue to the cap before the cleanup runs, so the
+	// insert site must suppress instead of growing it.
+	for i := 0; i < adminnotify.MaxUnacknowledgedCriticalLogs; i++ {
+		if _, err := tx.Exec(ctx, `
+			INSERT INTO admin_notifications (reason, user_id, created_at)
+			VALUES ('deposit_not_paid_by_deadline', $1, NOW())
+		`, userID); err != nil {
+			t.Fatalf("failed to seed deposit_not_paid_by_deadline notification %d: %v", i, err)
+		}
+	}
+	if !adminnotify.CriticalLogsCapExceeded(ctx, db.Conn, "deposit_not_paid_by_deadline") {
+		t.Fatal("expected the unacknowledged deposit_not_paid_by_deadline queue to be at the cap")
+	}
+
+	_, err = CleanupExpiredDeposits(ctx)
+	if err != nil {
+		t.Fatalf("CleanupExpiredDeposits failed: %v", err)
+	}
+
+	// The eviction + time_blocker cleanup still run; only the notification
+	// insert is suppressed by the cap.
+	var status string
+	if err := tx.QueryRow(ctx, "SELECT status FROM bookings WHERE id = $1", bookingID).Scan(&status); err != nil {
+		t.Fatalf("failed to query booking: %v", err)
+	}
+	if status != "pending_release" {
+		t.Errorf("expected status 'pending_release', got '%s'", status)
+	}
+
+	var tbCount int
+	if err := tx.QueryRow(ctx, "SELECT COUNT(*) FROM time_blockers WHERE description = $1", "RESERVATION:user:"+userID+":bk123").Scan(&tbCount); err != nil {
+		t.Fatalf("failed to query time blockers: %v", err)
+	}
+	if tbCount != 0 {
+		t.Error("expected reservation time blocker to be deleted")
+	}
+
+	var n int
+	if err := tx.QueryRow(ctx, `SELECT COUNT(*) FROM admin_notifications WHERE reason = 'deposit_not_paid_by_deadline'`).Scan(&n); err != nil {
+		t.Fatalf("failed to count deposit_not_paid_by_deadline notifications: %v", err)
+	}
+	if n != adminnotify.MaxUnacknowledgedCriticalLogs {
+		t.Errorf("expected the queue to stay capped at %d, got %d", adminnotify.MaxUnacknowledgedCriticalLogs, n)
+	}
+}

backend/internal/adminnotify/adminnotify.go

@@ -16,18 +16,13 @@
 //     (dispute, booking, unknown-event, orphan-replay).
 //   - handlers/user/account.go InsertSquareErasureCriticalNotification.
 //
-// Sites owned by OTHER agents that still need the fold (coordination notes):
+// Sites owned by OTHER agents (coordination notes):
 //
-//   - handlers/payments/sweep.go:1556 insertCriticalPaymentNotification (money
+//   - handlers/payments/sweep.go insertCriticalPaymentNotification — NOW capped
-//     agent) — its INSERT ... SELECT ... WHERE NOT EXISTS is the same
+//     (same atomic fold + pre-check as every other site).
-//     unbounded-across-accounts shape; fold `AND (SELECT COUNT(*) FROM
+//   - handlers/scheduling/time-blockers.go insertSquareCleanupCriticalNotification
-//     admin_notifications _an WHERE _an.reason = 'critical_payment_log' AND
+//     and internal/jobs/cleanup.go ScanCriticalPaymentLogs — both NOW capped
-//     _an.acknowledged_at IS NULL) < $N` (N = MaxUnacknowledgedCriticalLogs)
+//     (same atomic fold + pre-check as every other site).
-//     into its WHERE clause.
-//   - handlers/scheduling/time-blockers.go:511 insertSquareCleanupCriticalNotification
-//     and internal/jobs/cleanup.go:333 ScanCriticalPaymentLogs — the same
-//     'critical_payment_log' insert shape on the GDPR-cleanup and log-scan
-//     paths (owner: the jobs/scheduling agent).
 //   - main.go has NO admin_notifications insert sites (it only mounts the
 //     notification read/ack routes), so nothing to cap there.
 //

backend/internal/adminnotify/adminnotify_test.go

@@ -0,0 +1,72 @@
+//go:build test
+
+package adminnotify
+
+import (
+	"context"
+	"os"
+	"testing"
+
+	"crussell/db"
+	"crussell/testutils/testdb"
+)
+
+func TestMain(m *testing.M) {
+	pool := testdb.CreateTestDatabase("crussell_test_adminnotify")
+	db.Conn = db.NewPoolProxy(pool)
+	code := m.Run()
+	testdb.DestroyTestDatabase(pool, "crussell_test_adminnotify")
+	os.Exit(code)
+}
+
+// TestMaxUnacknowledgedCriticalLogs_Value pins the exported flood-cap constant
+// that every capped insert site folds into its SQL (changing it re-scopes every
+// site at once, so the value is part of the shared contract).
+func TestMaxUnacknowledgedCriticalLogs_Value(t *testing.T) {
+	if MaxUnacknowledgedCriticalLogs != 100 {
+		t.Errorf("expected MaxUnacknowledgedCriticalLogs to be 100, got %d", MaxUnacknowledgedCriticalLogs)
+	}
+}
+
+// TestCriticalLogsCapExceeded verifies the exported pre-check that logs a
+// suppression: below the cap it reports false, at the cap true, and
+// acknowledging rows re-arms it.
+func TestCriticalLogsCapExceeded(t *testing.T) {
+	ctx := context.Background()
+	t.Cleanup(func() {
+		_, _ = db.Conn.Exec(ctx, "DELETE FROM admin_notifications WHERE reason = 'critical_payment_log'")
+	})
+
+	if CriticalLogsCapExceeded(ctx, db.Conn, "critical_payment_log") {
+		t.Fatal("expected the empty queue to be below the cap")
+	}
+
+	for i := 0; i < MaxUnacknowledgedCriticalLogs-1; i++ {
+		if _, err := db.Conn.Exec(ctx, `
+			INSERT INTO admin_notifications (reason, created_at)
+			VALUES ('critical_payment_log', NOW())
+		`); err != nil {
+			t.Fatalf("failed to insert row %d: %v", i, err)
+		}
+	}
+	if CriticalLogsCapExceeded(ctx, db.Conn, "critical_payment_log") {
+		t.Fatalf("expected %d unacknowledged rows to stay below the cap", MaxUnacknowledgedCriticalLogs-1)
+	}
+
+	if _, err := db.Conn.Exec(ctx, `
+		INSERT INTO admin_notifications (reason, created_at)
+		VALUES ('critical_payment_log', NOW())
+	`); err != nil {
+		t.Fatalf("failed to insert the cap row: %v", err)
+	}
+	if !CriticalLogsCapExceeded(ctx, db.Conn, "critical_payment_log") {
+		t.Fatalf("expected %d unacknowledged rows to be at the cap", MaxUnacknowledgedCriticalLogs)
+	}
+
+	if _, err := db.Conn.Exec(ctx, "UPDATE admin_notifications SET acknowledged_at = NOW() WHERE reason = 'critical_payment_log'"); err != nil {
+		t.Fatalf("failed to acknowledge the queue: %v", err)
+	}
+	if CriticalLogsCapExceeded(ctx, db.Conn, "critical_payment_log") {
+		t.Fatal("expected acknowledging the queue to re-arm inserts")
+	}
+}

backend/internal/jobs/cleanup.go

@@ -15,6 +15,7 @@ import (
 	"crussell/handlers/payments"
 	"crussell/handlers/scheduling"
 	"crussell/handlers/user"
+	"crussell/internal/adminnotify"
 	"crussell/internal/square"
 	"crussell/mw"
 
@@ -324,10 +325,28 @@ const criticalPaymentStaleAge = "2 hours"
 // the bell never floods. Acknowledging the notification re-arms the scan:
 // while the row stays unresolved it is surfaced again on the next run.
 //
+// Flood cap (Round 2 Loop B finding 1): the unacknowledged
+// 'critical_payment_log' queue is globally capped at
+// adminnotify.MaxUnacknowledgedCriticalLogs — the SAME shared cap as every
+// other insert site (webhooks, account erasure, time blockers). The cap is
+// folded into the INSERT's WHERE clause (count-then-insert is atomic, closing
+// the TOCTOU where two concurrent scans could both read a below-cap count and
+// overshoot together), and the pre-check logs the suppression so a capped-out
+// scan stays visible to the operator. A hostile flood of unresolved money rows
+// (attacker-registered accounts, a runaway reconciliation loop) must not be
+// able to bury the single-operator notification centre.
+//
 // This is the "grep CRITICAL" the audit asked for, but DB-backed since the app
 // has no log pipeline. When a proper log/alert pipeline (T14 Sentry) lands,
 // this job can be retired.
 func ScanCriticalPaymentLogs(ctx context.Context) (int, error) {
+	// Pre-check: at the cap, skip the scan entirely (the fold inside the INSERT
+	// enforces the cap atomically even if a concurrent insert races this check;
+	// the pre-check only decides whether to log the suppression).
+	if adminnotify.CriticalLogsCapExceeded(ctx, db.Conn, "critical_payment_log") {
+		log.Printf("[SCAN] critical_payment_log admin notification suppressed — %d unacknowledged rows at the cap; acknowledge outstanding notifications to re-arm", adminnotify.MaxUnacknowledgedCriticalLogs)
+		return 0, nil
+	}
 	tag, err := db.Conn.Exec(ctx, `
 		INSERT INTO admin_notifications (reason, booking_id, created_at)
 		SELECT DISTINCT 'critical_payment_log'::admin_notification_reason, src.booking_id, NOW()
@@ -358,7 +377,14 @@ func ScanCriticalPaymentLogs(ctx context.Context) (int, error) {
 		AND (src.booking_id IS NULL OR EXISTS (
 			SELECT 1 FROM bookings b WHERE b.id = src.booking_id
 		))
-	`)
+		-- Flood cap (Round 2 Loop B finding 1): fold the global
+		-- MaxUnacknowledgedCriticalLogs bound INTO the insert so the
+		-- count-then-insert is atomic — two concurrent scans cannot both read a
+		-- below-cap count and overshoot together.
+		AND (SELECT COUNT(*) FROM admin_notifications _an
+		     WHERE _an.reason = 'critical_payment_log'
+		       AND _an.acknowledged_at IS NULL) < $1
+	`, adminnotify.MaxUnacknowledgedCriticalLogs)
 	if err != nil {
 		return 0, fmt.Errorf("failed to scan critical payment logs: %w", err)
 	}

backend/internal/jobs/cleanup_test.go

@@ -13,6 +13,7 @@ import (
 
 	"crussell/db"
 	"crussell/handlers/payments"
+	"crussell/internal/adminnotify"
 	"crussell/internal/square"
 	"crussell/testutils"
 	"crussell/testutils/testdb"
@@ -331,6 +332,60 @@ func TestScanCriticalPaymentLogs_RefundBelowCapNotNotified(t *testing.T) {
 	}
 }
 
+// TestScanCriticalPaymentLogs_FloodCapped verifies the shared
+// 'critical_payment_log' flood cap (Round 2 Loop B finding 1): once the
+// unacknowledged queue reaches adminnotify.MaxUnacknowledgedCriticalLogs, the
+// scan inserts nothing, and acknowledging rows re-arms it.
+func TestScanCriticalPaymentLogs_FloodCapped(t *testing.T) {
+	ctx := context.Background()
+	paymentID := seedStalePendingPayment(t)
+	t.Cleanup(func() {
+		_, _ = db.Conn.Exec(ctx, "DELETE FROM payments WHERE id = $1", paymentID)
+		_, _ = db.Conn.Exec(ctx, "DELETE FROM admin_notifications WHERE reason = 'critical_payment_log'")
+	})
+
+	// Fill the unacknowledged queue to the cap.
+	for i := 0; i < adminnotify.MaxUnacknowledgedCriticalLogs; i++ {
+		if _, err := db.Conn.Exec(ctx, `
+			INSERT INTO admin_notifications (reason, created_at)
+			VALUES ('critical_payment_log', NOW())
+		`); err != nil {
+			t.Fatalf("failed to fill the unacknowledged queue to the cap: %v", err)
+		}
+	}
+	if !adminnotify.CriticalLogsCapExceeded(ctx, db.Conn, "critical_payment_log") {
+		t.Fatal("expected the unacknowledged critical_payment_log queue to be at the cap")
+	}
+
+	// At the cap the scan must insert nothing for the stale pending payment.
+	n, err := ScanCriticalPaymentLogs(ctx)
+	if err != nil {
+		t.Fatalf("ScanCriticalPaymentLogs failed: %v", err)
+	}
+	if n != 0 {
+		t.Errorf("expected 0 notifications inserted at the cap, got %d", n)
+	}
+	var got int
+	if err := db.Conn.QueryRow(ctx, `SELECT COUNT(*) FROM admin_notifications WHERE reason = 'critical_payment_log' AND acknowledged_at IS NULL`).Scan(&got); err != nil {
+		t.Fatalf("failed to count unacknowledged notifications: %v", err)
+	}
+	if got != adminnotify.MaxUnacknowledgedCriticalLogs {
+		t.Errorf("expected the queue to stay capped at %d, got %d", adminnotify.MaxUnacknowledgedCriticalLogs, got)
+	}
+
+	// Acknowledging re-arms the scan while the row stays unresolved.
+	if _, err := db.Conn.Exec(ctx, "UPDATE admin_notifications SET acknowledged_at = NOW() WHERE reason = 'critical_payment_log' AND acknowledged_at IS NULL"); err != nil {
+		t.Fatalf("failed to acknowledge the queue: %v", err)
+	}
+	n, err = ScanCriticalPaymentLogs(ctx)
+	if err != nil {
+		t.Fatalf("ScanCriticalPaymentLogs failed after acknowledge: %v", err)
+	}
+	if n != 1 {
+		t.Errorf("expected 1 notification re-inserted after acknowledge, got %d", n)
+	}
+}
+
 // ============================================================
 // RetryPendingSquareErasures — GDPR Square outbox job (batch-1 fix)
 // ============================================================