Add web auth, infinite-scroll, agent admission and review fixes across agent/server/web

CONFIG_REFERENCE.ru.md

@@ -0,0 +1,206 @@
+# Monlet: краткий справочник по настройкам
+
+Monlet состоит из агента на хостах и центрального сервера. Агент сам генерирует локальный ключ в `state_dir/agent.key`; сервер видит новый ключ как `pending`, пока оператор не примет агента в UI.
+
+## Agent TOML
+
+Файл: обычно `/etc/monlet/agent.toml`.
+
+```toml
+# Необязательно. Если не задано, используется hostname ОС.
+# agent_id = "host-01"
+
+state_dir = "/var/lib/monlet-agent"
+
+[labels]
+env = "prod"
+role = "api"
+
+[server]
+enabled = true
+url = "https://monlet.example.com"
+heartbeat_interval = "10s"
+batch_interval = "10s"
+
+[metrics]
+enabled = true
+listen = "127.0.0.1:9465"
+
+[[checks]]
+id = "disk_root"
+name = "Root disk usage"
+command = '''
+/usr/local/lib/monlet/check_disk_root.sh --warn 80 --crit 90
+'''
+interval = "60s"
+timeout = "10s"
+notifications_enabled = true
+resource_limits = { cpu_time = "5s", memory = "256MiB", open_files = 128 }
+```
+
+Ключи:
+
+- `agent_id` — стабильный ID агента; необязателен, по умолчанию hostname.
+- `state_dir` — локальное состояние, ключ агента и spool.
+- `[labels]` — метки инвентаря; секреты сюда не класть.
+- `server.enabled` — отправлять heartbeat/events на сервер.
+- `server.url` — базовый URL сервера.
+- `metrics.enabled` — поднять Prometheus `/metrics`.
+- `metrics.listen` — адрес `/metrics`.
+- `checks[].id` — стабильный ID проверки.
+- `checks[].command` — shell-команда строкой; однострочно в `"..."`, многострочно в `'''...'''`.
+- `checks[].interval` — период запуска.
+- `checks[].timeout` — таймаут; при таймауте результат `critical`.
+- `checks[].notifications_enabled` — разрешить серверные уведомления по этой проверке.
+- `checks[].resource_limits` — опциональные лимиты процесса: CPU-время, virtual memory, open files.
+
+Exit code: `0=ok`, `1=warning`, `2=critical`, `3+=unknown`.
+
+Env агента:
+
+- `MONLET_AGENT_MAX_CHECKS` — максимум проверок в конфиге (по умолчанию 256). Лимит ограничивает кардинальность метрики `{check_id}`. Положительное целое; превышение лимита — ошибка загрузки конфига.
+
+Reload:
+
+- `systemctl reload monlet-agent` перечитывает локальный TOML через `SIGHUP`.
+- Ошибочный config не применяется, старый продолжает работать.
+- Уже запущенный check при reload не убивается, а доживает до своего `timeout`.
+- Для изменения `agent_id`, `state_dir`, `server.*`, `metrics.*` нужен restart.
+
+## Server env
+
+Файл: `.env` или переменные контейнера.
+
+```env
+MONLET_AUTH_TOKEN=replace-with-random-ui-token
+MONLET_DATABASE_URL=postgresql+asyncpg://monlet:monlet@postgres:5432/monlet
+MONLET_LOG_LEVEL=INFO
+
+MONLET_HOST=0.0.0.0
+MONLET_PORT=8000
+
+MONLET_BODY_LIMIT_BYTES=1048576
+MONLET_OUTPUT_MAX_BYTES=8192
+MONLET_MAX_BATCH_EVENTS=200
+
+MONLET_ENABLE_DETECTOR=true
+MONLET_STALE_AFTER_SEC=20
+MONLET_DEAD_AFTER_SEC=30
+MONLET_DETECTOR_TICK_SEC=5
+
+MONLET_DB_POOL_SIZE=10
+MONLET_DB_MAX_OVERFLOW=10
+MONLET_DB_POOL_TIMEOUT_SEC=30
+MONLET_DB_POOL_RECYCLE_SEC=1800
+
+MONLET_EVENTS_RETENTION_MONTHS=36
+MONLET_EVENTS_FUTURE_PARTITIONS=3
+MONLET_PARTITION_MAINTENANCE_INTERVAL_SEC=3600
+MONLET_OUTBOX_RETENTION_MAX_ROWS=50000
+
+MONLET_EVENT_DEDUP_RETENTION_DAYS=30
+MONLET_EVENT_DEDUP_PRUNE_BATCH_SIZE=5000
+
+MONLET_MAX_PENDING_AGENTS=10000
+MONLET_PENDING_AGENT_TTL_DAYS=7
+
+MONLET_ENABLE_NOTIFIER_WORKER=true
+MONLET_NOTIFIER_TICK_SEC=5
+MONLET_NOTIFIER_BATCH_SIZE=20
+MONLET_NOTIFIER_CONCURRENCY=4
+MONLET_NOTIFIER_MAX_ATTEMPTS=8
+MONLET_NOTIFIER_LEASE_SEC=300
+MONLET_NOTIFIER_HTTP_TIMEOUT_SEC=10
+MONLET_NOTIFIER_INCLUDE_OUTPUT=true
+MONLET_NOTIFIER_MAX_OUTPUT_BYTES=1024
+MONLET_NOTIFICATION_TIME_ZONE=UTC
+
+MONLET_NOTIFIER_DEBUG_ENABLED=true
+MONLET_NOTIFIER_TELEGRAM_ENABLED=false
+MONLET_NOTIFIER_WEBHOOK_ENABLED=false
+MONLET_NOTIFIER_ALERTMANAGER_ENABLED=false
+```
+
+Основное:
+
+- `MONLET_AUTH_TOKEN` — Bearer-токен(ы) для UI/operator API, не для агентов. Список через запятую/пробел для ротации.
+- `MONLET_DATABASE_URL` — PostgreSQL DSN (`postgresql+asyncpg://...`).
+- `MONLET_LOG_LEVEL` — уровень логирования.
+- `MONLET_HOST` / `MONLET_PORT` — адрес и порт HTTP-сервера (по умолчанию `0.0.0.0:8000`).
+
+Лимиты приёма:
+
+- `MONLET_BODY_LIMIT_BYTES` — максимальный размер тела запроса (по умолчанию 1 MiB = 1048576).
+- `MONLET_OUTPUT_MAX_BYTES` — лимит хранимого `output` события (по умолчанию 8192).
+- `MONLET_MAX_BATCH_EVENTS` — максимум событий в одном batch (по умолчанию 200).
+
+Liveness-детектор:
+
+- `MONLET_ENABLE_DETECTOR` — включить фоновый пересчёт stale/dead (по умолчанию `true`).
+- `MONLET_STALE_AFTER_SEC` — через сколько секунд без heartbeat агент становится `stale`.
+- `MONLET_DEAD_AFTER_SEC` — через сколько секунд без heartbeat агент становится `dead`.
+- `MONLET_DETECTOR_TICK_SEC` — как часто сервер пересчитывает liveness.
+
+Пул БД (тюнинг под нагрузку):
+
+- `MONLET_DB_POOL_SIZE` — постоянных соединений в пуле (по умолчанию 10).
+- `MONLET_DB_MAX_OVERFLOW` — дополнительных соединений сверх пула (по умолчанию 10).
+- `MONLET_DB_POOL_TIMEOUT_SEC` — ожидание свободного соединения (по умолчанию 30).
+- `MONLET_DB_POOL_RECYCLE_SEC` — пересоздание соединения по возрасту (по умолчанию 1800).
+
+Хранение и обслуживание:
+
+- `MONLET_EVENTS_RETENTION_MONTHS` — хранение event-партиций (по умолчанию 36).
+- `MONLET_EVENTS_FUTURE_PARTITIONS` — сколько будущих месячных партиций создавать заранее (по умолчанию 3).
+- `MONLET_PARTITION_MAINTENANCE_INTERVAL_SEC` — период обслуживания партиций и prune (по умолчанию 3600).
+- `MONLET_OUTBOX_RETENTION_MAX_ROWS` — верхняя граница строк в outbox; лишние обрезаются (по умолчанию 50000).
+- `MONLET_EVENT_DEDUP_RETENTION_DAYS` — хранение записей идемпотентности `event_id` (по умолчанию 30).
+- `MONLET_EVENT_DEDUP_PRUNE_BATCH_SIZE` — размер пакета при очистке dedup-записей (по умолчанию 5000).
+- `MONLET_MAX_PENDING_AGENTS` — лимит новых непринятых агентов (по умолчанию 10000).
+- `MONLET_PENDING_AGENT_TTL_DAYS` — очистка старых pending-ключей (по умолчанию 7).
+
+Notifier worker:
+
+- `MONLET_ENABLE_NOTIFIER_WORKER` — включить worker доставки уведомлений (по умолчанию `true`).
+- `MONLET_NOTIFIER_TICK_SEC` — период опроса outbox (по умолчанию 5).
+- `MONLET_NOTIFIER_BATCH_SIZE` — сколько outbox-записей worker забирает за тик (по умолчанию 20).
+- `MONLET_NOTIFIER_CONCURRENCY` — параллелизм доставки внутри тика (по умолчанию 4).
+- `MONLET_NOTIFIER_MAX_ATTEMPTS` — максимум попыток доставки до перевода в permanent failure (по умолчанию 8).
+- `MONLET_NOTIFIER_LEASE_SEC` — аренда записи на время доставки, чтобы её не взял другой worker (по умолчанию 300).
+- `MONLET_NOTIFIER_HTTP_TIMEOUT_SEC` — таймаут HTTP-доставки (по умолчанию 10).
+- `MONLET_NOTIFIER_INCLUDE_OUTPUT` — включать ли текст `output` в уведомления (по умолчанию `true`).
+- `MONLET_NOTIFIER_MAX_OUTPUT_BYTES` — обрезка `output` в уведомлении после редакции (по умолчанию 1024).
+- `MONLET_NOTIFICATION_TIME_ZONE` — таймзона человекочитаемых дат в уведомлениях.
+
+Notifier каналы:
+
+- Telegram: `MONLET_NOTIFIER_TELEGRAM_ENABLED=true`, `MONLET_NOTIFIER_TELEGRAM_TOKEN`, `MONLET_NOTIFIER_TELEGRAM_CHAT_ID`.
+- Webhook: `MONLET_NOTIFIER_WEBHOOK_ENABLED=true`, `MONLET_NOTIFIER_WEBHOOK_URL`, опционально `MONLET_NOTIFIER_WEBHOOK_TOKEN`.
+- Alertmanager: `MONLET_NOTIFIER_ALERTMANAGER_ENABLED=true`, `MONLET_NOTIFIER_ALERTMANAGER_URL`.
+- Debug/log: `MONLET_NOTIFIER_DEBUG_ENABLED=true`.
+
+## Web env
+
+Веб-UI читает auth-настройки из env. Без них read-only страницы доступны локально, но admission-мутации возвращают `403`.
+
+```env
+MONLET_WEB_AUTH_USERNAME=admin
+MONLET_WEB_AUTH_PASSWORD=replace-with-strong-password
+MONLET_WEB_SESSION_SECRET=replace-with-random-32-bytes-min
+MONLET_WEB_SESSION_TTL_SEC=604800
+# или вместо локального логина — доверие reverse-proxy:
+# MONLET_WEB_TRUST_PROXY_AUTH=true
+```
+
+- `MONLET_WEB_AUTH_USERNAME` / `MONLET_WEB_AUTH_PASSWORD` — учётные данные локального cookie-логина.
+- `MONLET_WEB_SESSION_SECRET` — секрет подписи сессии, минимум 32 байта; короче — UI считает auth неправильно настроенным.
+- `MONLET_WEB_SESSION_TTL_SEC` — срок жизни сессии (по умолчанию 604800 = 7 дней).
+- `MONLET_WEB_TRUST_PROXY_AUTH` — `true` включает доверие заголовку `X-Forwarded-User` от reverse-proxy. Несовместимо с локальным логином: одновременная настройка обоих режимов — ошибка (UI отвечает `500`).
+
+## Минимальная схема
+
+1. Сервер стартует с PostgreSQL и `MONLET_AUTH_TOKEN`.
+2. Агент стартует с TOML, создаёт `state_dir/agent.key` и шлёт heartbeat.
+3. Новый агент появляется в UI как `pending`.
+4. Оператор принимает агента.
+5. После accept сервер начинает учитывать events, checks, incidents и notifications.

agent/README.md

@@ -26,6 +26,7 @@ See `config.example.toml`. Required fields: top-level `state_dir`, at least one
 | `metrics.listen` | `127.0.0.1:9465` | low-cardinality only |
 | `checks[].command` | required | Shell string executed as `/bin/sh -c`. Both forms accepted: `command = "check_disk --warn 80 --crit 90"` and triple-quoted `command = '''...'''`. Argv arrays are rejected. See [Command security contract](#command-security-contract). |
 | `checks[].notifications_enabled` | `true` | Set `false` for checks that must not create server notifications. |
+| `checks[].resource_limits` | `{}` | Optional per-check best-effort `ulimit` values: `cpu_time`, `memory`, `open_files`. |
 
 Examples — one-line and multi-line forms:
 
@@ -44,6 +45,7 @@ set -eu
 '''
 interval = "60s"
 timeout = "10s"
+resource_limits = { cpu_time = "5s", memory = "256MiB", open_files = 128 }
 ```
 
 ## Command security contract
@@ -67,9 +69,25 @@ executes it as `/bin/sh -c <command>` on the host where the agent runs.
   validation to keep secrets out of heartbeats.
 - Each run uses `exec.CommandContext` with `Setpgid`. On timeout the whole
   process group is killed and the check result becomes `critical`.
+- Optional `resource_limits` are applied inside the child shell before the
+  configured command runs. `cpu_time` uses `ulimit -t` and maps SIGXCPU to
+  `critical`; `memory` uses virtual memory/address-space `ulimit -v`; `open_files`
+  uses `ulimit -n`. These limits are best-effort and are not a cgroup/RSS limit.
 - Combined stdout+stderr is truncated by UTF-8 byte length to 8 KiB with marker
   `...[truncated N bytes]` before being persisted, exposed, or notified.
 
+## Config reload
+
+Send `SIGHUP` or run `systemctl reload monlet-agent` to reload the local TOML file.
+
+- Reload first validates the full new config. On error, the old config keeps running.
+- Hot-reloadable: labels and checks, including `command`, `interval`, `timeout`,
+  `notifications_enabled`, `dedupe_key`, and `resource_limits`.
+- Restart required: `agent_id`, `state_dir`, `server.*`, and `metrics.*`.
+- A check already running during reload is allowed to finish under its old config.
+  Removed or changed checks do not start new old-config runs; changed checks start
+  with the new config after the in-flight run finishes.
+
 ## Behaviour
 
 - Per-check scheduler with no-overlap: a tick is skipped if the previous run is still in flight (`monlet_agent_check_skipped_total`).
@@ -79,7 +97,9 @@ executes it as `/bin/sh -c <command>` on the host where the agent runs.
 - `output` (stdout+stderr) is truncated by **UTF-8 byte length** to 8 KiB with marker `...[truncated N bytes]`.
 - `event_id` is UUIDv7 generated on check completion and preserved across spool replay (ADR-0006 idempotency).
 - Spool: per-event JSON files in `state_dir/spool/`, FIFO, limits 10 000 events / 50 MiB, drop oldest on overflow (ADR-0005). Drops are counted in `monlet_agent_events_dropped_total{reason}` with bounded reasons; one warning is logged per ~30s drop burst, not per event.
+- Event flushes are split before POST so each JSON request body stays within the 1 MiB server limit.
 - Retry backoff for heartbeat and events: 1s → 30s, exp ×2, jitter ±20%. 4xx are dropped (not retried) to avoid hot loop; dropped batches are counted as `reason=send_non_retryable` and logged once per batch.
+- On shutdown the agent asks workers to stop cleanly; if they do not drain within 30s, the process exits non-zero so the service manager can restart or mark failure.
 - Bearer auth header on both endpoints uses the local `state_dir/agent.key` value (ADR-0007).
 - Heartbeats include configured labels plus reserved `monlet_agent_version=<binary version>`.
 - Heartbeats include explicit feature flags: `push` from `server.enabled`, `metrics` from `metrics.enabled`.
@@ -97,6 +117,11 @@ Exposed on `metrics.listen` at `/metrics`. Labels are static (`check_id`, `statu
 - `monlet_agent_check_last_success_timestamp_seconds{check_id}`
 - `monlet_agent_check_interval_seconds{check_id}`
 - `monlet_agent_checks_configured`
+- `monlet_agent_config_load_success`
+- `monlet_agent_config_reload_total{result}`
+- `monlet_agent_config_last_reload_success_timestamp_seconds`
+- `monlet_agent_resource_limit_apply_failures_total{check_id,resource}`
+- `monlet_agent_events_channel_full_total`
 - `monlet_agent_spool_events`, `monlet_agent_spool_bytes`
 - `monlet_agent_send_attempts_total{kind}`, `monlet_agent_send_failures_total{kind}` (`kind` ∈ `heartbeat`, `events`)
 - `monlet_agent_events_accepted_total`, `monlet_agent_events_deduplicated_total`

agent/cmd/monlet-agent/main.go

@@ -8,6 +8,7 @@ import (
 	"os/signal"
 	"path/filepath"
 	"syscall"
+	"time"
 
 	"github.com/monlet/agent/internal/agentkey"
 	"github.com/monlet/agent/internal/app"
@@ -20,6 +21,13 @@ import (
 
 var version = "0.1.0"
 
+const shutdownTimeout = 30 * time.Second
+
+type reloadTarget interface {
+	RecordReloadFailure()
+	Reload(*config.Config, string) error
+}
+
 func main() {
 	cfgPath := flag.String("config", "config.toml", "path to TOML config")
 	flag.Parse()
@@ -65,6 +73,12 @@ func main() {
 
 	ctx, cancel := signal.NotifyContext(context.Background(), syscall.SIGINT, syscall.SIGTERM)
 	defer cancel()
+	startShutdownWatchdog(ctx, shutdownTimeout, log, os.Exit)
+
+	hup := make(chan os.Signal, 1)
+	signal.Notify(hup, syscall.SIGHUP)
+	defer signal.Stop(hup)
+	go runReloadLoop(ctx, *cfgPath, hup, a, log)
 
 	log.Info("monlet-agent starting", "agent_id", agentID, "version", version, "checks", len(cfg.Checks))
 	if err := a.Run(ctx); err != nil {
@@ -73,3 +87,45 @@ func main() {
 	}
 	log.Info("monlet-agent stopped")
 }
+
+func startShutdownWatchdog(ctx context.Context, timeout time.Duration, log *slog.Logger, exit func(int)) {
+	go func() {
+		<-ctx.Done()
+		t := time.NewTimer(timeout)
+		defer t.Stop()
+		<-t.C
+		log.Error("shutdown timeout exceeded", "timeout", timeout.String())
+		exit(1)
+	}()
+}
+
+func runReloadLoop(ctx context.Context, cfgPath string, hup <-chan os.Signal, target reloadTarget, log *slog.Logger) {
+	for {
+		select {
+		case <-ctx.Done():
+			return
+		case <-hup:
+			reloadOnce(cfgPath, target, log)
+		}
+	}
+}
+
+func reloadOnce(cfgPath string, target reloadTarget, log *slog.Logger) {
+	nextCfg, err := config.Load(cfgPath)
+	if err != nil {
+		target.RecordReloadFailure()
+		log.Warn("reload config failed", "err", err)
+		return
+	}
+	nextAgentID, err := ids.ResolveAgentID(nextCfg.AgentID, nextCfg.Hostname)
+	if err != nil {
+		target.RecordReloadFailure()
+		log.Warn("reload agent_id failed", "err", err)
+		return
+	}
+	if err := target.Reload(nextCfg, nextAgentID); err != nil {
+		log.Warn("reload rejected", "err", err)
+		return
+	}
+	log.Info("reload complete", "checks", len(nextCfg.Checks))
+}

agent/cmd/monlet-agent/main_test.go

@@ -0,0 +1,118 @@
+package main
+
+import (
+	"context"
+	"fmt"
+	"io"
+	"log/slog"
+	"os"
+	"path/filepath"
+	"syscall"
+	"testing"
+	"time"
+
+	"github.com/monlet/agent/internal/config"
+)
+
+type reloadCall struct {
+	cfg     *config.Config
+	agentID string
+}
+
+type fakeReloadTarget struct {
+	reloads  chan reloadCall
+	failures chan struct{}
+}
+
+func (f *fakeReloadTarget) RecordReloadFailure() {
+	f.failures <- struct{}{}
+}
+
+func (f *fakeReloadTarget) Reload(cfg *config.Config, agentID string) error {
+	f.reloads <- reloadCall{cfg: cfg, agentID: agentID}
+	return nil
+}
+
+func TestRunReloadLoopHandlesSIGHUP(t *testing.T) {
+	cfgPath := writeTestConfig(t)
+	target := &fakeReloadTarget{
+		reloads:  make(chan reloadCall, 1),
+		failures: make(chan struct{}, 1),
+	}
+	ctx, cancel := context.WithCancel(context.Background())
+	defer cancel()
+	hup := make(chan os.Signal, 1)
+	go runReloadLoop(ctx, cfgPath, hup, target, discardLogger())
+
+	hup <- syscall.SIGHUP
+	select {
+	case call := <-target.reloads:
+		if call.agentID != "agent-test" || len(call.cfg.Checks) != 1 {
+			t.Fatalf("unexpected reload call: agent_id=%q cfg=%+v", call.agentID, call.cfg)
+		}
+	case <-time.After(time.Second):
+		t.Fatal("reload was not called")
+	}
+}
+
+func TestRunReloadLoopRecordsLoadFailure(t *testing.T) {
+	target := &fakeReloadTarget{
+		reloads:  make(chan reloadCall, 1),
+		failures: make(chan struct{}, 1),
+	}
+	ctx, cancel := context.WithCancel(context.Background())
+	defer cancel()
+	hup := make(chan os.Signal, 1)
+	go runReloadLoop(ctx, filepath.Join(t.TempDir(), "missing.toml"), hup, target, discardLogger())
+
+	hup <- syscall.SIGHUP
+	select {
+	case <-target.failures:
+	case <-time.After(time.Second):
+		t.Fatal("reload failure was not recorded")
+	}
+}
+
+func TestShutdownWatchdogExitsAfterTimeout(t *testing.T) {
+	ctx, cancel := context.WithCancel(context.Background())
+	exits := make(chan int, 1)
+	startShutdownWatchdog(ctx, time.Millisecond, discardLogger(), func(code int) {
+		exits <- code
+	})
+	cancel()
+	select {
+	case code := <-exits:
+		if code != 1 {
+			t.Fatalf("exit code: %d", code)
+		}
+	case <-time.After(time.Second):
+		t.Fatal("watchdog did not exit")
+	}
+}
+
+func writeTestConfig(t *testing.T) string {
+	t.Helper()
+	path := filepath.Join(t.TempDir(), "agent.toml")
+	body := fmt.Sprintf(`
+agent_id = "agent-test"
+state_dir = %q
+
+[server]
+enabled = true
+url = "http://localhost"
+
+[[checks]]
+id = "c1"
+command = "true"
+interval = "10s"
+timeout = "5s"
+`, t.TempDir())
+	if err := os.WriteFile(path, []byte(body), 0o600); err != nil {
+		t.Fatal(err)
+	}
+	return path
+}
+
+func discardLogger() *slog.Logger {
+	return slog.New(slog.NewTextHandler(io.Discard, nil))
+}

agent/config.example.toml

@@ -24,6 +24,7 @@ command = '''
 '''
 interval = "60s"
 timeout = "10s"
+resource_limits = { cpu_time = "5s", memory = "256MiB", open_files = 128 }
 
 [[checks]]
 id = "uptime"

agent/internal/app/app.go

@@ -3,6 +3,7 @@ package app
 import (
 	"context"
 	"errors"
+	"fmt"
 	"log/slog"
 	"sync"
 	"time"
@@ -24,6 +25,9 @@ type App struct {
 	Log     *slog.Logger
 
 	lastDropLog time.Time
+	cfgMu       sync.RWMutex
+	schedulerMu sync.RWMutex
+	scheduler   *scheduler.Manager
 }
 
 const (
@@ -36,31 +40,30 @@ const (
 
 func (a *App) Run(ctx context.Context) error {
 	events := make(chan scheduler.Event, 256)
+	cfg := a.config()
 
 	// Account drops accumulated during spool Open (e.g. shrunk limits on restart).
 	a.recordSpoolDrops(a.Spool.Drops())
-	a.Metrics.ChecksConfigured.Set(float64(len(a.Cfg.Checks)))
+	a.Metrics.ConfigLoadSuccess.Set(1)
-	a.Metrics.BuildInfo.WithLabelValues(a.Version, boolLabel(a.Cfg.PushesToServer()), boolLabel(a.Cfg.ExposesMetrics())).Set(1)
+	a.Metrics.ChecksConfigured.Set(float64(len(cfg.Checks)))
-	for _, c := range a.Cfg.Checks {
+	a.Metrics.BuildInfo.WithLabelValues(a.Version, boolLabel(cfg.PushesToServer()), boolLabel(cfg.ExposesMetrics())).Set(1)
+	for _, c := range cfg.Checks {
 		a.Metrics.CheckInterval.WithLabelValues(c.ID).Set(c.Interval.Duration.Seconds())
 	}
 
 	var wg sync.WaitGroup
 
-	if a.Cfg.ExposesMetrics() {
+	if cfg.ExposesMetrics() {
 		wg.Add(1)
 		go func() {
 			defer wg.Done()
-			if err := a.Metrics.Serve(ctx, a.Cfg.Metrics.Listen); err != nil {
+			if err := a.Metrics.Serve(ctx, cfg.Metrics.Listen); err != nil {
 				a.Log.Error("metrics server", "err", err)
 			}
 		}()
 	}
 
-	wg.Add(1)
+	mgr := scheduler.NewManager(ctx, a.AgentID, events, scheduler.Hooks{
-	go func() {
-		defer wg.Done()
-		scheduler.Run(ctx, a.AgentID, a.Cfg.Checks, events, scheduler.Hooks{
 		OnSkipped: func(id string) { a.Metrics.CheckSkipped.WithLabelValues(id).Inc() },
 		OnResult: func(ev scheduler.Event) {
 			a.Metrics.CheckRuns.WithLabelValues(ev.CheckID, ev.Status).Inc()
@@ -73,10 +76,25 @@ func (a *App) Run(ctx context.Context) error {
 				a.Metrics.CheckLastSuccess.WithLabelValues(ev.CheckID).Set(ts)
 			}
 		},
+		OnEventChannelFull: func() { a.Metrics.EventsChannelFull.Inc() },
+		OnResourceLimitFailure: func(id string, resource string) {
+			a.Metrics.ResourceLimitApplyFailures.WithLabelValues(id, resource).Inc()
+		},
+		OnStopped: func(id string) { a.Metrics.DeleteCheck(id) },
 	})
+	mgr.Update(cfg.Checks)
+	a.schedulerMu.Lock()
+	a.scheduler = mgr
+	a.schedulerMu.Unlock()
+	wg.Add(1)
+	go func() {
+		defer wg.Done()
+		<-ctx.Done()
+		mgr.Stop()
+		mgr.Wait()
 	}()
 
-	if a.Cfg.PushesToServer() {
+	if cfg.PushesToServer() {
 		wg.Add(1)
 		go func() {
 			defer wg.Done()
@@ -128,18 +146,19 @@ func (a *App) spoolWriter(ctx context.Context, events <-chan scheduler.Event) {
 }
 
 func (a *App) heartbeatLoop(ctx context.Context) {
-	interval := a.Cfg.Server.HeartbeatInterval.Duration
 	attempt := 0
 	for {
+		cfg := a.config()
+		interval := cfg.Server.HeartbeatInterval.Duration
 		hb := client.HeartbeatRequest{
 			AgentID:    a.AgentID,
 			ObservedAt: time.Now().UTC(),
-			Hostname:   a.Cfg.Hostname,
+			Hostname:   cfg.Hostname,
 			Features: client.AgentFeatures{
-				Push:    a.Cfg.PushesToServer(),
+				Push:    cfg.PushesToServer(),
-				Metrics: a.Cfg.ExposesMetrics(),
+				Metrics: cfg.ExposesMetrics(),
 			},
-			Labels: a.Cfg.HeartbeatLabels(a.Version),
+			Labels: cfg.HeartbeatLabels(a.Version),
 		}
 		a.Metrics.SendAttempts.WithLabelValues("heartbeat").Inc()
 		err := a.Client.SendHeartbeat(ctx, hb)
@@ -167,9 +186,14 @@ func (a *App) heartbeatLoop(ctx context.Context) {
 }
 
 func (a *App) sendLoop(ctx context.Context) {
-	interval := a.Cfg.Server.BatchInterval.Duration
 	attempt := 0
 	for {
+		select {
+		case <-ctx.Done():
+			return
+		default:
+		}
+		interval := a.config().Server.BatchInterval.Duration
 		items, err := a.Spool.Peek(client.MaxBatchEvents)
 		if err != nil {
 			a.Log.Error("spool peek", "err", err)
@@ -217,6 +241,74 @@ func (a *App) sendLoop(ctx context.Context) {
 	}
 }
 
+// Reload applies a validated hot-reload config if immutable fields match.
+func (a *App) Reload(newCfg *config.Config, newAgentID string) error {
+	oldCfg := a.config()
+	if err := validateReload(oldCfg, newCfg, a.AgentID, newAgentID); err != nil {
+		a.RecordReloadFailure()
+		return err
+	}
+	a.cfgMu.Lock()
+	a.Cfg = newCfg
+	a.cfgMu.Unlock()
+
+	a.Metrics.ConfigLoadSuccess.Set(1)
+	a.Metrics.ConfigReloadTotal.WithLabelValues("success").Inc()
+	a.Metrics.ConfigLastReloadSuccess.Set(float64(time.Now().Unix()))
+	a.Metrics.ChecksConfigured.Set(float64(len(newCfg.Checks)))
+	for _, c := range newCfg.Checks {
+		a.Metrics.CheckInterval.WithLabelValues(c.ID).Set(c.Interval.Duration.Seconds())
+	}
+
+	a.schedulerMu.RLock()
+	mgr := a.scheduler
+	a.schedulerMu.RUnlock()
+	if mgr != nil {
+		mgr.Update(newCfg.Checks)
+	}
+	return nil
+}
+
+// RecordReloadFailure updates config reload metrics after a failed reload attempt.
+func (a *App) RecordReloadFailure() {
+	a.Metrics.ConfigLoadSuccess.Set(0)
+	a.Metrics.ConfigReloadTotal.WithLabelValues("failure").Inc()
+}
+
+func (a *App) config() *config.Config {
+	a.cfgMu.RLock()
+	defer a.cfgMu.RUnlock()
+	return a.Cfg
+}
+
+func validateReload(oldCfg, newCfg *config.Config, oldAgentID, newAgentID string) error {
+	if oldAgentID != newAgentID {
+		return fmt.Errorf("agent_id change requires restart")
+	}
+	if oldCfg.StateDir != newCfg.StateDir {
+		return fmt.Errorf("state_dir change requires restart")
+	}
+	if oldCfg.PushesToServer() != newCfg.PushesToServer() {
+		return fmt.Errorf("server.enabled change requires restart")
+	}
+	if oldCfg.Server.URL != newCfg.Server.URL {
+		return fmt.Errorf("server.url change requires restart")
+	}
+	if oldCfg.Server.HeartbeatInterval.Duration != newCfg.Server.HeartbeatInterval.Duration {
+		return fmt.Errorf("server.heartbeat_interval change requires restart")
+	}
+	if oldCfg.Server.BatchInterval.Duration != newCfg.Server.BatchInterval.Duration {
+		return fmt.Errorf("server.batch_interval change requires restart")
+	}
+	if oldCfg.ExposesMetrics() != newCfg.ExposesMetrics() {
+		return fmt.Errorf("metrics.enabled change requires restart")
+	}
+	if oldCfg.Metrics.Listen != newCfg.Metrics.Listen {
+		return fmt.Errorf("metrics.listen change requires restart")
+	}
+	return nil
+}
+
 // sleep returns true if context was cancelled while waiting.
 func sleep(ctx context.Context, d time.Duration) bool {
 	t := time.NewTimer(d)

agent/internal/app/app_test.go

@@ -182,6 +182,81 @@ func TestEndToEndPushAndReplay(t *testing.T) {
 	}
 }
 
+func TestReloadRejectsImmutableChangesAndKeepsOldConfig(t *testing.T) {
+	oldCfg := &config.Config{
+		AgentID:  "a1",
+		Hostname: "h",
+		StateDir: "/tmp/old",
+		Server: config.ServerConfig{
+			Enabled:           boolPtr(true),
+			URL:               "http://old",
+			HeartbeatInterval: config.Duration{Duration: 10 * time.Second},
+			BatchInterval:     config.Duration{Duration: 10 * time.Second},
+		},
+		Metrics: config.MetricsConfig{Enabled: false},
+		Checks: []config.CheckConfig{{
+			ID:       "c1",
+			Command:  "true",
+			Interval: config.Duration{Duration: time.Second},
+			Timeout:  config.Duration{Duration: time.Second},
+		}},
+	}
+	newCfg := *oldCfg
+	newCfg.Server.URL = "http://new"
+	a := &App{
+		Cfg:     oldCfg,
+		AgentID: "a1",
+		Metrics: metrics.New(),
+	}
+	if err := a.Reload(&newCfg, "a1"); err == nil {
+		t.Fatal("expected immutable reload rejection")
+	}
+	if a.config().Server.URL != "http://old" {
+		t.Fatalf("old config was replaced: %+v", a.config().Server)
+	}
+}
+
+func TestReloadAcceptsLabelsAndChecks(t *testing.T) {
+	oldCfg := &config.Config{
+		AgentID:  "a1",
+		Hostname: "h",
+		StateDir: "/tmp/old",
+		Labels:   map[string]string{"env": "old"},
+		Server: config.ServerConfig{
+			Enabled:           boolPtr(true),
+			URL:               "http://server",
+			HeartbeatInterval: config.Duration{Duration: 10 * time.Second},
+			BatchInterval:     config.Duration{Duration: 10 * time.Second},
+		},
+		Metrics: config.MetricsConfig{Enabled: false},
+		Checks: []config.CheckConfig{{
+			ID:       "c1",
+			Command:  "true",
+			Interval: config.Duration{Duration: time.Second},
+			Timeout:  config.Duration{Duration: time.Second},
+		}},
+	}
+	newCfg := *oldCfg
+	newCfg.Labels = map[string]string{"env": "new"}
+	newCfg.Checks = []config.CheckConfig{{
+		ID:       "c2",
+		Command:  "true",
+		Interval: config.Duration{Duration: 2 * time.Second},
+		Timeout:  config.Duration{Duration: time.Second},
+	}}
+	a := &App{
+		Cfg:     oldCfg,
+		AgentID: "a1",
+		Metrics: metrics.New(),
+	}
+	if err := a.Reload(&newCfg, "a1"); err != nil {
+		t.Fatal(err)
+	}
+	if a.config().Labels["env"] != "new" || len(a.config().Checks) != 1 || a.config().Checks[0].ID != "c2" {
+		t.Fatalf("reload did not apply: %+v", a.config())
+	}
+}
+
 func boolPtr(v bool) *bool {
 	return &v
 }

agent/internal/client/client.go

@@ -4,6 +4,7 @@ import (
 	"bytes"
 	"context"
 	"encoding/json"
+	"errors"
 	"fmt"
 	"io"
 	"net/http"
@@ -15,6 +16,7 @@ import (
 
 const (
 	MaxBatchEvents = 200
+	MaxBatchBytes  = 1 << 20
 )
 
 type Client struct {
@@ -68,18 +70,81 @@ func (c *Client) SendEvents(ctx context.Context, events []scheduler.Event) (*Eve
 		return &EventBatchResponse{}, nil
 	}
 	if len(events) > MaxBatchEvents {
-		return nil, fmt.Errorf("batch exceeds %d events", MaxBatchEvents)
+		return nil, nonRetryablef("batch exceeds %d events", MaxBatchEvents)
 	}
 	wireEvents := make([]scheduler.Event, len(events))
 	for i, ev := range events {
 		wireEvents[i] = ev.WireEvent()
 	}
 	req := EventBatchRequest{AgentID: c.agentID, Events: wireEvents}
-	var resp EventBatchResponse
+	batches, err := c.splitEventBatches(req.Events)
-	if err := c.post(ctx, "/api/v1/events", req, &resp); err != nil {
+	if err != nil {
 		return nil, err
 	}
-	return &resp, nil
+	var total EventBatchResponse
+	for _, batch := range batches {
+		var resp EventBatchResponse
+		if err := c.post(ctx, "/api/v1/events", EventBatchRequest{AgentID: c.agentID, Events: batch}, &resp); err != nil {
+			return nil, err
+		}
+		total.Accepted += resp.Accepted
+		total.Deduplicated += resp.Deduplicated
+	}
+	return &total, nil
+}
+
+func (c *Client) splitEventBatches(events []scheduler.Event) ([][]scheduler.Event, error) {
+	overhead, err := c.eventBatchOverhead()
+	if err != nil {
+		return nil, err
+	}
+	batches := make([][]scheduler.Event, 0, 1)
+	current := make([]scheduler.Event, 0, len(events))
+	currentSize := overhead
+	for _, ev := range events {
+		eventSize, err := jsonSize(ev)
+		if err != nil {
+			return nil, err
+		}
+		nextSize := currentSize + eventSize
+		if len(current) > 0 {
+			nextSize++
+		}
+		if nextSize <= MaxBatchBytes {
+			current = append(current, ev)
+			currentSize = nextSize
+			continue
+		}
+		if len(current) == 0 {
+			return nil, nonRetryablef("single event batch exceeds %d bytes", MaxBatchBytes)
+		}
+		batches = append(batches, current)
+		current = []scheduler.Event{ev}
+		currentSize = overhead + eventSize
+		if currentSize > MaxBatchBytes {
+			return nil, nonRetryablef("single event batch exceeds %d bytes", MaxBatchBytes)
+		}
+	}
+	if len(current) > 0 {
+		batches = append(batches, current)
+	}
+	return batches, nil
+}
+
+func (c *Client) eventBatchOverhead() (int, error) {
+	size, err := jsonSize(EventBatchRequest{AgentID: c.agentID, Events: []scheduler.Event{}})
+	if err != nil {
+		return 0, err
+	}
+	return size, nil
+}
+
+func jsonSize(v any) (int, error) {
+	b, err := json.Marshal(v)
+	if err != nil {
+		return 0, err
+	}
+	return len(b), nil
 }
 
 func (c *Client) post(ctx context.Context, path string, body any, out any) error {
@@ -115,30 +180,28 @@ type HTTPError struct {
 
 func (e *HTTPError) Error() string { return fmt.Sprintf("http %d: %s", e.Status, e.Body) }
 
+type NonRetryableError struct{ Err error }
+
+func (e *NonRetryableError) Error() string { return e.Err.Error() }
+func (e *NonRetryableError) Unwrap() error { return e.Err }
+
+func nonRetryablef(format string, args ...any) error {
+	return &NonRetryableError{Err: fmt.Errorf(format, args...)}
+}
+
 // Retryable reports whether the error should trigger a retry (network errors
 // and 5xx are retryable; 4xx are dropped to avoid hot-loop on bad payload).
 func Retryable(err error) bool {
 	if err == nil {
 		return false
 	}
+	var ne *NonRetryableError
+	if errors.As(err, &ne) {
+		return false
+	}
 	var he *HTTPError
-	if asErr(err, &he) {
+	if errors.As(err, &he) {
 		return he.Status >= 500
 	}
 	return true
 }
-
-func asErr(err error, target **HTTPError) bool {
-	for e := err; e != nil; {
-		if h, ok := e.(*HTTPError); ok {
-			*target = h
-			return true
-		}
-		u, ok := e.(interface{ Unwrap() error })
-		if !ok {
-			return false
-		}
-		e = u.Unwrap()
-	}
-	return false
-}

agent/internal/client/client_test.go

@@ -6,6 +6,7 @@ import (
 	"io"
 	"net/http"
 	"net/http/httptest"
+	"strings"
 	"testing"
 	"time"
 
@@ -57,6 +58,59 @@ func TestSendEventsBatch(t *testing.T) {
 	}
 }
 
+func TestSendEventsSplitsByJSONSize(t *testing.T) {
+	var (
+		requests int
+		total    int
+	)
+	srv := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		body, err := io.ReadAll(r.Body)
+		if err != nil {
+			t.Error(err)
+			http.Error(w, "read body", http.StatusInternalServerError)
+			return
+		}
+		if len(body) > MaxBatchBytes {
+			t.Errorf("body size %d exceeds %d", len(body), MaxBatchBytes)
+			http.Error(w, "too large", http.StatusRequestEntityTooLarge)
+			return
+		}
+		var got EventBatchRequest
+		if err := json.Unmarshal(body, &got); err != nil {
+			t.Error(err)
+			http.Error(w, "bad json", http.StatusBadRequest)
+			return
+		}
+		requests++
+		total += len(got.Events)
+		w.WriteHeader(http.StatusAccepted)
+		_ = json.NewEncoder(w).Encode(EventBatchResponse{Accepted: len(got.Events)})
+	}))
+	defer srv.Close()
+
+	events := make([]scheduler.Event, MaxBatchEvents)
+	for i := range events {
+		events[i] = scheduler.Event{
+			EventID:    "event-" + time.Unix(int64(i), 0).UTC().Format("20060102150405"),
+			CheckID:    "check",
+			ObservedAt: time.Unix(int64(i), 0).UTC(),
+			Status:     "ok",
+			Output:     strings.Repeat("x", 8192),
+		}
+	}
+	c := New(srv.URL, "tok", "a1")
+	resp, err := c.SendEvents(context.Background(), events)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if requests < 2 {
+		t.Fatalf("expected split batch, got %d request", requests)
+	}
+	if total != MaxBatchEvents || resp.Accepted != MaxBatchEvents {
+		t.Fatalf("total=%d resp=%+v", total, resp)
+	}
+}
+
 func TestSendEventsTooBig(t *testing.T) {
 	c := New("http://x", "t", "a1")
 	big := make([]scheduler.Event, MaxBatchEvents+1)
@@ -64,6 +118,24 @@ func TestSendEventsTooBig(t *testing.T) {
 	if err == nil {
 		t.Fatal("expected error")
 	}
+	if Retryable(err) {
+		t.Fatal("oversized batch should not retry")
+	}
+}
+
+func TestSendEventsSingleEventTooLargeIsNotRetryable(t *testing.T) {
+	c := New("http://x", "t", "a1")
+	_, err := c.SendEvents(context.Background(), []scheduler.Event{{
+		EventID: "x",
+		CheckID: strings.Repeat("c", MaxBatchBytes),
+		Status:  "ok",
+	}})
+	if err == nil {
+		t.Fatal("expected error")
+	}
+	if Retryable(err) {
+		t.Fatal("oversized single event should not retry")
+	}
 }
 
 func TestRetryableClassification(t *testing.T) {

agent/internal/config/config.go

@@ -2,6 +2,7 @@ package config
 
 import (
 	"fmt"
+	"math"
 	"os"
 	"regexp"
 	"strconv"
@@ -60,10 +61,21 @@ type CheckConfig struct {
 	Timeout              Duration       `toml:"timeout"`
 	NotificationsEnabled *bool          `toml:"notifications_enabled"`
 	DedupeKey            string         `toml:"dedupe_key"`
+	ResourceLimits       ResourceLimits `toml:"resource_limits"`
 }
 
 type Duration struct{ time.Duration }
 
+// ByteSize stores a parsed TOML byte size.
+type ByteSize struct{ Bytes int64 }
+
+// ResourceLimits defines optional per-check OS resource limits.
+type ResourceLimits struct {
+	CPUTime   Duration `toml:"cpu_time"`
+	Memory    ByteSize `toml:"memory"`
+	OpenFiles int      `toml:"open_files"`
+}
+
 func (d *Duration) UnmarshalText(b []byte) error {
 	v, err := time.ParseDuration(string(b))
 	if err != nil {
@@ -73,6 +85,43 @@ func (d *Duration) UnmarshalText(b []byte) error {
 	return nil
 }
 
+func (s *ByteSize) UnmarshalText(b []byte) error {
+	raw := strings.TrimSpace(string(b))
+	if raw == "" {
+		return fmt.Errorf("empty byte size")
+	}
+	valueEnd := 0
+	for valueEnd < len(raw) && ((raw[valueEnd] >= '0' && raw[valueEnd] <= '9') || raw[valueEnd] == '.') {
+		valueEnd++
+	}
+	if valueEnd == 0 {
+		return fmt.Errorf("invalid byte size %q", raw)
+	}
+	value, err := strconv.ParseFloat(raw[:valueEnd], 64)
+	if err != nil || value <= 0 {
+		return fmt.Errorf("invalid byte size %q", raw)
+	}
+	unit := strings.ToLower(strings.TrimSpace(raw[valueEnd:]))
+	mul, ok := map[string]float64{
+		"": 1, "b": 1,
+		"k": 1000, "kb": 1000,
+		"m": 1000 * 1000, "mb": 1000 * 1000,
+		"g": 1000 * 1000 * 1000, "gb": 1000 * 1000 * 1000,
+		"kib": 1024,
+		"mib": 1024 * 1024,
+		"gib": 1024 * 1024 * 1024,
+	}[unit]
+	if !ok {
+		return fmt.Errorf("unknown byte size unit %q", unit)
+	}
+	bytes := value * mul
+	if bytes > math.MaxInt64 {
+		return fmt.Errorf("byte size %q is too large", raw)
+	}
+	s.Bytes = int64(bytes)
+	return nil
+}
+
 func Load(path string) (*Config, error) {
 	var c Config
 	md, err := toml.DecodeFile(path, &c)
@@ -174,6 +223,9 @@ func (c *Config) Validate() error {
 		if len(ch.DedupeKey) > 256 {
 			return fmt.Errorf("check %q: dedupe_key too long", ch.ID)
 		}
+		if err := validateResourceLimits(ch.ID, ch.ResourceLimits); err != nil {
+			return err
+		}
 	}
 	return nil
 }
@@ -201,8 +253,23 @@ func (c CheckConfig) NotificationsOn() bool {
 	return c.NotificationsEnabled == nil || *c.NotificationsEnabled
 }
 
-func (c CheckConfig) Argv() []string {
+func validateResourceLimits(checkID string, limits ResourceLimits) error {
-	return []string{"/bin/sh", "-c", c.Command}
+	if limits.CPUTime.Duration < 0 {
+		return fmt.Errorf("check %q: resource_limits.cpu_time must be > 0", checkID)
+	}
+	if limits.Memory.Bytes < 0 {
+		return fmt.Errorf("check %q: resource_limits.memory must be > 0", checkID)
+	}
+	if limits.Memory.Bytes > 0 && limits.Memory.Bytes < 1024*1024 {
+		return fmt.Errorf("check %q: resource_limits.memory must be >= 1MiB", checkID)
+	}
+	if limits.OpenFiles < 0 {
+		return fmt.Errorf("check %q: resource_limits.open_files must be > 0", checkID)
+	}
+	if limits.OpenFiles > 0 && limits.OpenFiles < 16 {
+		return fmt.Errorf("check %q: resource_limits.open_files must be >= 16", checkID)
+	}
+	return nil
 }
 
 func ValidateID(field, v string) error {

agent/internal/config/config_test.go

@@ -48,6 +48,59 @@ func TestLoadMinimal(t *testing.T) {
 	}
 }
 
+func TestLoadResourceLimits(t *testing.T) {
+	c, err := Load(writeTmp(t, `
+agent_id = "host-1"
+state_dir = "/tmp/x"
+
+[server]
+enabled = true
+url = "http://localhost"
+
+[[checks]]
+id = "c1"
+command = "true"
+interval = "10s"
+timeout = "5s"
+resource_limits = { cpu_time = "2s", memory = "256MiB", open_files = 128 }
+`))
+	if err != nil {
+		t.Fatal(err)
+	}
+	limits := c.Checks[0].ResourceLimits
+	if limits.CPUTime.Duration != 2_000_000_000 || limits.Memory.Bytes != 256*1024*1024 || limits.OpenFiles != 128 {
+		t.Fatalf("unexpected limits: %+v", limits)
+	}
+}
+
+func TestValidateBadResourceLimits(t *testing.T) {
+	for name, line := range map[string]string{
+		"memory_too_small":     `resource_limits = { memory = "512KiB" }`,
+		"open_files_too_small": `resource_limits = { open_files = 8 }`,
+	} {
+		t.Run(name, func(t *testing.T) {
+			_, err := Load(writeTmp(t, `
+agent_id = "host-1"
+state_dir = "/tmp/x"
+
+[server]
+enabled = true
+url = "http://localhost"
+
+[[checks]]
+id = "c1"
+command = "true"
+interval = "10s"
+timeout = "5s"
+`+line+`
+`))
+			if err == nil {
+				t.Fatal("expected resource limit error")
+			}
+		})
+	}
+}
+
 func TestLoadMultilineCommand(t *testing.T) {
 	c, err := Load(writeTmp(t, `
 agent_id = "host-1"
@@ -72,9 +125,6 @@ timeout = "5s"
 	if !strings.Contains(c.Checks[0].Command, "echo ok") {
 		t.Fatalf("command not loaded: %q", c.Checks[0].Command)
 	}
-	if got := c.Checks[0].Argv(); len(got) != 3 || got[0] != "/bin/sh" || got[1] != "-c" {
-		t.Fatalf("argv: %#v", got)
-	}
 }
 
 func TestCommandArrayRejected(t *testing.T) {

agent/internal/metrics/metrics.go

@@ -22,6 +22,11 @@ type Metrics struct {
 	CheckLastSuccess           *prometheus.GaugeVec
 	CheckInterval              *prometheus.GaugeVec
 	ChecksConfigured           prometheus.Gauge
+	ConfigLoadSuccess          prometheus.Gauge
+	ConfigReloadTotal          *prometheus.CounterVec
+	ConfigLastReloadSuccess    prometheus.Gauge
+	ResourceLimitApplyFailures *prometheus.CounterVec
+	EventsChannelFull          prometheus.Counter
 	SpoolDepth                 prometheus.Gauge
 	SpoolBytes                 prometheus.Gauge
 	SendAttempts               *prometheus.CounterVec
@@ -115,6 +120,26 @@ func New() *Metrics {
 		Name: "monlet_agent_checks_configured",
 		Help: "Number of checks configured.",
 	})
+	m.ConfigLoadSuccess = prometheus.NewGauge(prometheus.GaugeOpts{
+		Name: "monlet_agent_config_load_success",
+		Help: "Whether the last agent config load or reload succeeded.",
+	})
+	m.ConfigReloadTotal = prometheus.NewCounterVec(prometheus.CounterOpts{
+		Name: "monlet_agent_config_reload_total",
+		Help: "Number of config reload attempts by result.",
+	}, []string{"result"})
+	m.ConfigLastReloadSuccess = prometheus.NewGauge(prometheus.GaugeOpts{
+		Name: "monlet_agent_config_last_reload_success_timestamp_seconds",
+		Help: "Unix timestamp of the last successful config reload.",
+	})
+	m.ResourceLimitApplyFailures = prometheus.NewCounterVec(prometheus.CounterOpts{
+		Name: "monlet_agent_resource_limit_apply_failures_total",
+		Help: "Number of failed resource limit applications by check and resource.",
+	}, []string{"check_id", "resource"})
+	m.EventsChannelFull = prometheus.NewCounter(prometheus.CounterOpts{
+		Name: "monlet_agent_events_channel_full_total",
+		Help: "Number of times scheduler event delivery would block because the agent event channel was full.",
+	})
 	m.LastHeartbeat = prometheus.NewGauge(prometheus.GaugeOpts{
 		Name: "monlet_agent_last_heartbeat_timestamp_seconds",
 		Help: "Unix timestamp of the last successful heartbeat ack.",
@@ -126,6 +151,9 @@ func New() *Metrics {
 	reg.MustRegister(m.CheckRuns, m.CheckSkipped, m.CheckDuration,
 		m.CheckStatus, m.CheckExitCode, m.CheckLastRun, m.CheckLastSuccess,
 		m.CheckInterval, m.ChecksConfigured,
+		m.ConfigLoadSuccess, m.ConfigReloadTotal, m.ConfigLastReloadSuccess,
+		m.ResourceLimitApplyFailures,
+		m.EventsChannelFull,
 		m.SpoolDepth, m.SpoolBytes,
 		m.SendAttempts, m.SendFailures,
 		m.EventsAccepted, m.EventsDedup, m.EventsDropped,
@@ -133,6 +161,19 @@ func New() *Metrics {
 	return m
 }
 
+// DeleteCheck removes metric series for a check that no longer exists.
+func (m *Metrics) DeleteCheck(checkID string) {
+	m.CheckRuns.DeletePartialMatch(prometheus.Labels{"check_id": checkID})
+	m.CheckSkipped.DeleteLabelValues(checkID)
+	m.CheckDuration.DeleteLabelValues(checkID)
+	m.CheckStatus.DeleteLabelValues(checkID)
+	m.CheckExitCode.DeleteLabelValues(checkID)
+	m.CheckLastRun.DeleteLabelValues(checkID)
+	m.CheckLastSuccess.DeleteLabelValues(checkID)
+	m.CheckInterval.DeleteLabelValues(checkID)
+	m.ResourceLimitApplyFailures.DeletePartialMatch(prometheus.Labels{"check_id": checkID})
+}
+
 // Serve blocks until ctx is done; the listener is closed on shutdown.
 func (m *Metrics) Serve(ctx context.Context, addr string) error {
 	mux := http.NewServeMux()

agent/internal/metrics/metrics_test.go

@@ -14,6 +14,7 @@ func TestMetricsExposeRegistered(t *testing.T) {
 	m := New()
 	m.CheckRuns.WithLabelValues("c1", "ok").Inc()
 	m.SpoolDepth.Set(7)
+	m.EventsChannelFull.Inc()
 	srv := httptest.NewServer(promhttp.HandlerFor(m.Registry, promhttp.HandlerOpts{}))
 	defer srv.Close()
 	resp, err := http.Get(srv.URL)
@@ -26,6 +27,7 @@ func TestMetricsExposeRegistered(t *testing.T) {
 	for _, want := range []string{
 		`monlet_agent_check_runs_total{check_id="c1",status="ok"} 1`,
 		`monlet_agent_spool_events 7`,
+		`monlet_agent_events_channel_full_total 1`,
 	} {
 		if !strings.Contains(s, want) {
 			t.Errorf("missing %q in output", want)

agent/internal/runner/runner.go

@@ -6,6 +6,8 @@ import (
 	"errors"
 	"fmt"
 	"os/exec"
+	"strconv"
+	"strings"
 	"syscall"
 	"time"
 	"unicode/utf8"
@@ -16,6 +18,7 @@ var replacementRune = []byte("<22>")
 const (
 	MaxOutputBytes  = 8192
 	truncMarkerFmt  = "...[truncated %d bytes]"
+	limitFailPrefix = "__MONLET_RESOURCE_LIMIT_FAILED__:"
 )
 
 type Status string
@@ -33,11 +36,36 @@ type Result struct {
 	DurationMs           int64
 	Output               string
 	OutputTruncated      bool
+	ResourceLimitFailure string
+}
+
+// ResourceLimits contains best-effort shell ulimit values for a check command.
+type ResourceLimits struct {
+	CPUTime     time.Duration
+	MemoryBytes int64
+	OpenFiles   int
+}
+
+func (l ResourceLimits) IsZero() bool {
+	return l.CPUTime == 0 && l.MemoryBytes == 0 && l.OpenFiles == 0
 }
 
 // Run executes argv with timeout. Stdout+stderr are captured into a single
 // buffer; both are truncated to 8 KiB UTF-8 bytes if exceeded.
 func Run(ctx context.Context, argv []string, timeout time.Duration) Result {
+	return runArgv(ctx, argv, timeout)
+}
+
+// RunCommand executes a shell command with optional resource limits.
+func RunCommand(ctx context.Context, command string, timeout time.Duration, limits ResourceLimits) Result {
+	argv := []string{"/bin/sh", "-c", command}
+	if !limits.IsZero() {
+		argv = limitedShellArgv(command, limits)
+	}
+	return runArgv(ctx, argv, timeout)
+}
+
+func runArgv(ctx context.Context, argv []string, timeout time.Duration) Result {
 	start := time.Now()
 	cctx, cancel := context.WithTimeout(ctx, timeout)
 	defer cancel()
@@ -67,6 +95,23 @@ func Run(ctx context.Context, argv []string, timeout time.Duration) Result {
 		r.Output = fmt.Sprintf("critical: check timed out after %s", timeout)
 		return r
 	}
+	if isExitCode(err, 125) {
+		if resource, ok := detectLimitFailure(out); ok {
+			r.Status = StatusUnknown
+			r.ExitCode = -1
+			r.Output = "unknown: failed to apply resource limit: " + resource
+			r.OutputTruncated = false
+			r.ResourceLimitFailure = resource
+			return r
+		}
+	}
+	if isSignal(err, syscall.SIGXCPU) {
+		r.Status = StatusCritical
+		r.ExitCode = 2
+		r.Output = "critical: CPU time limit exceeded"
+		r.OutputTruncated = false
+		return r
+	}
 	if err != nil {
 		var ee *exec.ExitError
 		if errors.As(err, &ee) {
@@ -74,6 +119,8 @@ func Run(ctx context.Context, argv []string, timeout time.Duration) Result {
 		} else {
 			r.Status = StatusUnknown
 			r.ExitCode = -1
+			// Launch failure (e.g. shell missing); surface exec error, never check output.
+			r.Output, r.OutputTruncated = TruncateUTF8([]byte("unknown: failed to start check: "+err.Error()), MaxOutputBytes)
 			return r
 		}
 	}
@@ -81,6 +128,60 @@ func Run(ctx context.Context, argv []string, timeout time.Duration) Result {
 	return r
 }
 
+func isExitCode(err error, code int) bool {
+	var ee *exec.ExitError
+	return errors.As(err, &ee) && ee.ExitCode() == code
+}
+
+func limitedShellArgv(command string, limits ResourceLimits) []string {
+	cpu := ""
+	if limits.CPUTime > 0 {
+		cpu = strconv.FormatInt(int64((limits.CPUTime+time.Second-1)/time.Second), 10)
+	}
+	mem := ""
+	if limits.MemoryBytes > 0 {
+		mem = strconv.FormatInt((limits.MemoryBytes+1023)/1024, 10)
+	}
+	openFiles := ""
+	if limits.OpenFiles > 0 {
+		openFiles = strconv.Itoa(limits.OpenFiles)
+	}
+	script := `
+cmd=$1
+cpu=$2
+mem=$3
+open_files=$4
+if [ -n "$cpu" ]; then ulimit -t "$cpu" 2>/dev/null || { printf '%s\n' '` + limitFailPrefix + `cpu_time'; exit 125; }; fi
+if [ -n "$mem" ]; then ulimit -v "$mem" 2>/dev/null || { printf '%s\n' '` + limitFailPrefix + `memory'; exit 125; }; fi
+if [ -n "$open_files" ]; then ulimit -n "$open_files" 2>/dev/null || { printf '%s\n' '` + limitFailPrefix + `open_files'; exit 125; }; fi
+exec /bin/sh -c "$cmd"
+`
+	return []string{"/bin/sh", "-c", script, "monlet-limit-wrapper", command, cpu, mem, openFiles}
+}
+
+func detectLimitFailure(output string) (string, bool) {
+	for _, line := range strings.Split(output, "\n") {
+		line = strings.TrimSpace(line)
+		if strings.HasPrefix(line, limitFailPrefix) {
+			resource := strings.TrimPrefix(line, limitFailPrefix)
+			if resource != "cpu_time" && resource != "memory" && resource != "open_files" {
+				resource = "unknown"
+			}
+			return resource, true
+		}
+	}
+	return "", false
+}
+
+func isSignal(err error, sig syscall.Signal) bool {
+	var ee *exec.ExitError
+	if !errors.As(err, &ee) {
+		return false
+	}
+	ws, ok := ee.Sys().(syscall.WaitStatus)
+	return ok && ws.Signaled() && ws.Signal() == sig
+}
+
 func MapExit(code int) Status {
 	switch code {
 	case 0:

agent/internal/runner/runner_test.go

@@ -2,6 +2,7 @@ package runner
 
 import (
 	"context"
+	"runtime"
 	"strings"
 	"testing"
 	"time"
@@ -114,3 +115,41 @@ func TestRunNoExec(t *testing.T) {
 		t.Fatalf("got %+v", r)
 	}
 }
+
+func TestRunCommandWithLimitsOK(t *testing.T) {
+	r := RunCommand(context.Background(), "echo ok", 2*time.Second, ResourceLimits{
+		CPUTime:   time.Second,
+		OpenFiles: 64,
+	})
+	if r.Status != StatusOK || r.ExitCode != 0 {
+		t.Fatalf("got %+v", r)
+	}
+	if !strings.Contains(r.Output, "ok") {
+		t.Fatalf("output: %q", r.Output)
+	}
+}
+
+func TestRunCommandCPULimitExceeded(t *testing.T) {
+	if runtime.GOOS == "windows" {
+		t.Skip("ulimit is Unix-only")
+	}
+	r := RunCommand(context.Background(), "while :; do :; done", 5*time.Second, ResourceLimits{
+		CPUTime: time.Second,
+	})
+	if r.Status != StatusCritical || r.ExitCode != 2 {
+		t.Fatalf("got %+v", r)
+	}
+	if !strings.Contains(r.Output, "CPU time limit exceeded") {
+		t.Fatalf("output: %q", r.Output)
+	}
+}
+
+func TestDetectLimitFailureBoundsResourceLabel(t *testing.T) {
+	resource, ok := detectLimitFailure(limitFailPrefix + "dynamic-user-output")
+	if !ok {
+		t.Fatal("expected limit failure")
+	}
+	if resource != "unknown" {
+		t.Fatalf("resource label must be bounded, got %q", resource)
+	}
+}

agent/internal/scheduler/scheduler.go

@@ -2,6 +2,7 @@ package scheduler
 
 import (
 	"context"
+	"reflect"
 	"sync"
 	"time"
 
@@ -35,46 +36,284 @@ func (e Event) WireEvent() Event {
 type Hooks struct {
 	OnSkipped              func(checkID string)
 	OnResult               func(ev Event)
+	OnEventChannelFull     func()
+	OnResourceLimitFailure func(checkID string, resource string)
+	OnStopped              func(checkID string)
 }
 
-// Run starts one goroutine per check. Emits events to out. Blocks until ctx is done.
+// Manager owns per-check workers and applies hot config updates.
-func Run(ctx context.Context, agentID string, checks []config.CheckConfig, out chan<- Event, h Hooks) {
+type Manager struct {
-	var wg sync.WaitGroup
+	ctx     context.Context
-	for i := range checks {
+	agentID string
-		c := checks[i]
+	out     chan<- Event
-		wg.Add(1)
+	hooks   Hooks
-		go func() {
+
-			defer wg.Done()
+	mu       sync.Mutex
-			runCheck(ctx, agentID, c, out, h)
+	workers  map[string]*checkWorker
+	draining map[string]struct{}
+	pending  map[string]config.CheckConfig
+	all      []*checkWorker
+	stopped  bool
+}
+
+// NewManager creates a scheduler manager bound to the parent context.
+func NewManager(ctx context.Context, agentID string, out chan<- Event, h Hooks) *Manager {
+	return &Manager{
+		ctx:      ctx,
+		agentID:  agentID,
+		out:      out,
+		hooks:    h,
+		workers:  make(map[string]*checkWorker),
+		draining: make(map[string]struct{}),
+		pending:  make(map[string]config.CheckConfig),
+	}
+}
+
+// Update applies the current check set without killing in-flight runs.
+func (m *Manager) Update(checks []config.CheckConfig) {
+	m.mu.Lock()
+	defer m.mu.Unlock()
+	if m.stopped {
+		return
+	}
+	next := make(map[string]config.CheckConfig, len(checks))
+	for _, c := range checks {
+		next[c.ID] = c
+		if w, ok := m.workers[c.ID]; ok {
+			if _, draining := m.draining[c.ID]; draining {
+				m.pending[c.ID] = c
+			} else {
+				w.Update(c)
+			}
+			continue
+		}
+		id := c.ID
+		w := newCheckWorker(m.ctx, m.agentID, c, m.out, m.hooks, func(stopped *checkWorker) {
+			m.workerStopped(id, stopped)
+		})
+		m.workers[c.ID] = w
+		m.all = append(m.all, w)
+	}
+	for id, w := range m.workers {
+		if _, ok := next[id]; ok {
+			continue
+		}
+		m.draining[id] = struct{}{}
+		delete(m.pending, id)
+		w.Stop()
+	}
+}
+
+// Stop prevents new runs and lets in-flight checks finish or observe ctx cancel.
+func (m *Manager) Stop() {
+	m.mu.Lock()
+	defer m.mu.Unlock()
+	if m.stopped {
+		return
+	}
+	m.stopped = true
+	for _, w := range m.workers {
+		w.Stop()
+	}
+}
+
+// Wait blocks until all workers have drained.
+func (m *Manager) Wait() {
+	m.mu.Lock()
+	workers := append([]*checkWorker(nil), m.all...)
+	m.mu.Unlock()
+	for _, w := range workers {
+		w.Wait()
+	}
+}
+
+func (m *Manager) workerStopped(checkID string, w *checkWorker) {
+	m.mu.Lock()
+	defer m.mu.Unlock()
+	if current, ok := m.workers[checkID]; !ok || current != w {
+		return
+	}
+	delete(m.workers, checkID)
+	delete(m.draining, checkID)
+	if m.stopped {
+		delete(m.pending, checkID)
+		return
+	}
+	if cfg, ok := m.pending[checkID]; ok {
+		delete(m.pending, checkID)
+		next := newCheckWorker(m.ctx, m.agentID, cfg, m.out, m.hooks, func(stopped *checkWorker) {
+			m.workerStopped(checkID, stopped)
+		})
+		m.workers[checkID] = next
+		m.all = append(m.all, next)
+	}
+}
+
+type checkWorker struct {
+	ctx     context.Context
+	agentID string
+	out     chan<- Event
+	hooks   Hooks
+	onDone  func(*checkWorker)
+
+	updateCh chan config.CheckConfig
+	stopCh   chan struct{}
+	doneCh   chan struct{}
+	stopOnce sync.Once
+	wg       sync.WaitGroup
+}
+
+func newCheckWorker(ctx context.Context, agentID string, c config.CheckConfig, out chan<- Event, h Hooks, onDone func(*checkWorker)) *checkWorker {
+	w := &checkWorker{
+		ctx:      ctx,
+		agentID:  agentID,
+		out:      out,
+		hooks:    h,
+		onDone:   onDone,
+		updateCh: make(chan config.CheckConfig, 1),
+		stopCh:   make(chan struct{}),
+		doneCh:   make(chan struct{}, 1),
+	}
+	w.wg.Add(1)
+	go w.loop(c)
+	return w
+}
+
+func (w *checkWorker) Update(c config.CheckConfig) {
+	select {
+	case w.updateCh <- c:
+	default:
+		select {
+		case <-w.updateCh:
+		default:
+		}
+		w.updateCh <- c
+	}
+}
+
+func (w *checkWorker) Stop() {
+	w.stopOnce.Do(func() { close(w.stopCh) })
+}
+
+func (w *checkWorker) Wait() {
+	w.wg.Wait()
+}
+
+func (w *checkWorker) loop(initial config.CheckConfig) {
+	defer func() {
+		if w.onDone != nil {
+			w.onDone(w)
+		}
+		w.wg.Done()
 	}()
-	}
+	cfg := initial
-	wg.Wait()
+	t := time.NewTicker(cfg.Interval.Duration)
-}
-
-func runCheck(ctx context.Context, agentID string, c config.CheckConfig, out chan<- Event, h Hooks) {
-	t := time.NewTicker(c.Interval.Duration)
 	defer t.Stop()
-	var running bool
-	var mu sync.Mutex
 
-	exec := func() {
+	running := false
-		mu.Lock()
+	stopping := false
+	runAfterCurrent := false
+	ctxDone := w.ctx.Done()
+	stopCh := w.stopCh
+
+	start := func(c config.CheckConfig) {
 		if running {
-			mu.Unlock()
+			if w.hooks.OnSkipped != nil {
-			if h.OnSkipped != nil {
+				w.hooks.OnSkipped(c.ID)
-				h.OnSkipped(c.ID)
 			}
 			return
 		}
 		running = true
-		mu.Unlock()
+		go func() {
-		defer func() {
+			w.run(c)
-			mu.Lock()
+			w.doneCh <- struct{}{}
-			running = false
-			mu.Unlock()
 		}()
+	}
 
-		res := runner.Run(ctx, c.Argv(), c.Timeout.Duration)
+	start(cfg)
+
+	for {
+		select {
+		case <-ctxDone:
+			stopping = true
+			ctxDone = nil
+			if !running {
+				if w.hooks.OnStopped != nil {
+					w.hooks.OnStopped(cfg.ID)
+				}
+				return
+			}
+		case <-stopCh:
+			stopping = true
+			stopCh = nil
+			if !running {
+				if w.hooks.OnStopped != nil {
+					w.hooks.OnStopped(cfg.ID)
+				}
+				return
+			}
+		case next := <-w.updateCh:
+			if reflect.DeepEqual(cfg, next) {
+				continue
+			}
+			cfg = next
+			t.Reset(cfg.Interval.Duration)
+			if running {
+				runAfterCurrent = true
+			} else {
+				start(cfg)
+			}
+		case <-w.doneCh:
+			running = false
+			if !stopping {
+				select {
+				case <-stopCh:
+					stopping = true
+					stopCh = nil
+				default:
+				}
+			}
+			if stopping {
+				if w.hooks.OnStopped != nil {
+					w.hooks.OnStopped(cfg.ID)
+				}
+				return
+			}
+			if runAfterCurrent {
+				runAfterCurrent = false
+				for {
+					select {
+					case next := <-w.updateCh:
+						if !reflect.DeepEqual(cfg, next) {
+							cfg = next
+							t.Reset(cfg.Interval.Duration)
+						}
+					default:
+						start(cfg)
+						goto nextLoop
+					}
+				}
+			nextLoop:
+				continue
+			}
+		case <-t.C:
+			if stopping {
+				continue
+			}
+			start(cfg)
+		}
+	}
+}
+
+func (w *checkWorker) run(c config.CheckConfig) {
+	res := runner.RunCommand(w.ctx, c.Command, c.Timeout.Duration, runner.ResourceLimits{
+		CPUTime:     c.ResourceLimits.CPUTime.Duration,
+		MemoryBytes: c.ResourceLimits.Memory.Bytes,
+		OpenFiles:   c.ResourceLimits.OpenFiles,
+	})
+	if res.ResourceLimitFailure != "" && w.hooks.OnResourceLimitFailure != nil {
+		w.hooks.OnResourceLimitFailure(c.ID, res.ResourceLimitFailure)
+	}
 	id, err := eventid.New()
 	if err != nil {
 		return
@@ -89,26 +328,22 @@ func runCheck(ctx context.Context, agentID string, c config.CheckConfig, out cha
 		Output:               res.Output,
 		OutputTruncated:      res.OutputTruncated,
 		NotificationsEnabled: boolPtr(c.NotificationsOn()),
-			IncidentKey:          incidentKey(agentID, c),
+		IncidentKey:          incidentKey(w.agentID, c),
 	}
-		if h.OnResult != nil {
+	if w.hooks.OnResult != nil {
-			h.OnResult(ev)
+		w.hooks.OnResult(ev)
 	}
 	select {
-		case out <- ev:
+	case w.out <- ev:
-		case <-ctx.Done():
+	case <-w.ctx.Done():
+	default:
+		if w.hooks.OnEventChannelFull != nil {
+			w.hooks.OnEventChannelFull()
 		}
-	}
-
-	// Fire first run immediately so smoke tests don't wait a full interval.
-	go exec()
-
-	for {
 		select {
-		case <-ctx.Done():
+		case w.out <- ev:
-			return
+		case <-w.ctx.Done():
-		case <-t.C:
+		case <-w.stopCh:
-			go exec()
 		}
 	}
 }

agent/internal/scheduler/scheduler_test.go

@@ -2,6 +2,7 @@ package scheduler
 
 import (
 	"context"
+	"strings"
 	"sync/atomic"
 	"testing"
 	"time"
@@ -9,6 +10,14 @@ import (
 	"github.com/monlet/agent/internal/config"
 )
 
+func runScheduler(ctx context.Context, agentID string, checks []config.CheckConfig, out chan<- Event, h Hooks) {
+	m := NewManager(ctx, agentID, out, h)
+	m.Update(checks)
+	<-ctx.Done()
+	m.Stop()
+	m.Wait()
+}
+
 func mkCheck(id, cmd string, interval, timeout time.Duration) config.CheckConfig {
 	return config.CheckConfig{
 		ID:       id,
@@ -22,7 +31,7 @@ func TestEmitsEvents(t *testing.T) {
 	ctx, cancel := context.WithTimeout(context.Background(), 500*time.Millisecond)
 	defer cancel()
 	out := make(chan Event, 10)
-	go Run(ctx, "a1", []config.CheckConfig{mkCheck("c1", "echo hi", 200*time.Millisecond, 100*time.Millisecond)}, out, Hooks{})
+	go runScheduler(ctx, "a1", []config.CheckConfig{mkCheck("c1", "echo hi", 200*time.Millisecond, 100*time.Millisecond)}, out, Hooks{})
 	select {
 	case ev := <-out:
 		if ev.CheckID != "c1" || ev.Status != "ok" {
@@ -43,9 +52,168 @@ func TestNoOverlapSkips(t *testing.T) {
 	var skipped int32
 	hooks := Hooks{OnSkipped: func(string) { atomic.AddInt32(&skipped, 1) }}
 	c := mkCheck("slow", "sleep 1", 100*time.Millisecond, 900*time.Millisecond)
-	go Run(ctx, "a1", []config.CheckConfig{c}, out, hooks)
+	go runScheduler(ctx, "a1", []config.CheckConfig{c}, out, hooks)
 	<-ctx.Done()
 	if atomic.LoadInt32(&skipped) == 0 {
 		t.Fatal("expected at least one skipped tick")
 	}
 }
+
+func TestReportsFullEventChannel(t *testing.T) {
+	ctx, cancel := context.WithTimeout(context.Background(), 2*time.Second)
+	defer cancel()
+	out := make(chan Event, 1)
+	out <- Event{EventID: "occupied"}
+	var full int32
+	m := NewManager(ctx, "a1", out, Hooks{
+		OnEventChannelFull: func() { atomic.AddInt32(&full, 1) },
+	})
+	m.Update([]config.CheckConfig{mkCheck("c1", "echo hi", time.Hour, time.Second)})
+
+	deadline := time.Now().Add(time.Second)
+	for time.Now().Before(deadline) {
+		if atomic.LoadInt32(&full) > 0 {
+			break
+		}
+		time.Sleep(10 * time.Millisecond)
+	}
+	if atomic.LoadInt32(&full) == 0 {
+		t.Fatal("expected full channel hook")
+	}
+	<-out
+	select {
+	case ev := <-out:
+		if ev.CheckID != "c1" {
+			t.Fatalf("unexpected event: %+v", ev)
+		}
+	case <-ctx.Done():
+		t.Fatal("blocked event was not delivered")
+	}
+	m.Stop()
+	m.Wait()
+}
+
+func TestManagerReloadChangedCheckWaitsForRunning(t *testing.T) {
+	ctx, cancel := context.WithTimeout(context.Background(), 2*time.Second)
+	defer cancel()
+	out := make(chan Event, 10)
+	m := NewManager(ctx, "a1", out, Hooks{})
+	m.Update([]config.CheckConfig{mkCheck("c1", "sleep 0.3; echo old", time.Second, time.Second)})
+	time.Sleep(50 * time.Millisecond)
+	m.Update([]config.CheckConfig{mkCheck("c1", "echo new", time.Second, time.Second)})
+
+	var first, second Event
+	select {
+	case first = <-out:
+	case <-ctx.Done():
+		t.Fatal("no first event")
+	}
+	select {
+	case second = <-out:
+	case <-ctx.Done():
+		t.Fatal("no second event")
+	}
+	m.Stop()
+	m.Wait()
+
+	if !strings.Contains(first.Output, "old") {
+		t.Fatalf("first output: %q", first.Output)
+	}
+	if !strings.Contains(second.Output, "new") {
+		t.Fatalf("second output: %q", second.Output)
+	}
+}
+
+func TestManagerReloadRemovedCheckLetsRunningFinish(t *testing.T) {
+	ctx, cancel := context.WithTimeout(context.Background(), 2*time.Second)
+	defer cancel()
+	out := make(chan Event, 10)
+	stopped := make(chan string, 1)
+	m := NewManager(ctx, "a1", out, Hooks{OnStopped: func(id string) { stopped <- id }})
+	m.Update([]config.CheckConfig{mkCheck("c1", "sleep 0.2; echo old", time.Second, time.Second)})
+	time.Sleep(50 * time.Millisecond)
+	m.Update(nil)
+
+	select {
+	case ev := <-out:
+		if !strings.Contains(ev.Output, "old") {
+			t.Fatalf("output: %q", ev.Output)
+		}
+	case <-ctx.Done():
+		t.Fatal("removed running check did not finish")
+	}
+	select {
+	case id := <-stopped:
+		if id != "c1" {
+			t.Fatalf("stopped id: %q", id)
+		}
+	case <-ctx.Done():
+		t.Fatal("removed check did not stop")
+	}
+	m.Stop()
+	m.Wait()
+}
+
+func TestManagerRemoveThenReaddSameCheckDoesNotOverlap(t *testing.T) {
+	ctx, cancel := context.WithTimeout(context.Background(), 3*time.Second)
+	defer cancel()
+	out := make(chan Event, 10)
+	m := NewManager(ctx, "a1", out, Hooks{})
+	m.Update([]config.CheckConfig{mkCheck("c1", "sleep 0.25; echo old", time.Second, time.Second)})
+	time.Sleep(50 * time.Millisecond)
+	m.Update(nil)
+	m.Update([]config.CheckConfig{mkCheck("c1", "echo new", time.Second, time.Second)})
+
+	var first, second Event
+	select {
+	case first = <-out:
+	case <-ctx.Done():
+		t.Fatal("no first event")
+	}
+	select {
+	case second = <-out:
+	case <-ctx.Done():
+		t.Fatal("no second event")
+	}
+	m.Stop()
+	m.Wait()
+
+	if !strings.Contains(first.Output, "old") {
+		t.Fatalf("first output: %q", first.Output)
+	}
+	if !strings.Contains(second.Output, "new") {
+		t.Fatalf("second output: %q", second.Output)
+	}
+}
+
+func TestManagerUsesLatestPendingReloadAfterRunningFinishes(t *testing.T) {
+	ctx, cancel := context.WithTimeout(context.Background(), 3*time.Second)
+	defer cancel()
+	out := make(chan Event, 10)
+	m := NewManager(ctx, "a1", out, Hooks{})
+	m.Update([]config.CheckConfig{mkCheck("c1", "sleep 0.25; echo old", time.Second, time.Second)})
+	time.Sleep(50 * time.Millisecond)
+	m.Update([]config.CheckConfig{mkCheck("c1", "echo intermediate", time.Second, time.Second)})
+	m.Update([]config.CheckConfig{mkCheck("c1", "echo latest", time.Second, time.Second)})
+
+	var first, second Event
+	select {
+	case first = <-out:
+	case <-ctx.Done():
+		t.Fatal("no first event")
+	}
+	select {
+	case second = <-out:
+	case <-ctx.Done():
+		t.Fatal("no second event")
+	}
+	m.Stop()
+	m.Wait()
+
+	if !strings.Contains(first.Output, "old") {
+		t.Fatalf("first output: %q", first.Output)
+	}
+	if strings.Contains(second.Output, "intermediate") || !strings.Contains(second.Output, "latest") {
+		t.Fatalf("second output: %q", second.Output)
+	}
+}

agent/systemd/monlet-agent.service.example

@@ -6,8 +6,10 @@ Wants=network-online.target
 [Service]
 Type=simple
 ExecStart=/usr/local/bin/monlet-agent -config /etc/monlet/agent.toml
+ExecReload=/bin/kill -HUP $MAINPID
 Restart=always
 RestartSec=5s
+TimeoutStopSec=35s
 User=monlet
 Group=monlet
 StateDirectory=monlet-agent

api/openapi.yaml

@@ -107,7 +107,9 @@ paths:
         "401":
           $ref: "#/components/responses/Unauthorized"
         "403":
-          description: Agent key is blocked.
+          description: >-
+            Agent key is blocked (`agent_blocked`) or not yet accepted
+            (`agent_not_accepted`).
           content:
             application/json:
               schema:
@@ -627,6 +629,7 @@ components:
                 - forbidden
                 - conflict
                 - agent_blocked
+                - agent_not_accepted
                 - not_found
                 - validation
                 - payload_too_large

deploy/docker-compose.yml

@@ -65,6 +65,8 @@ services:
       MONLET_WEB_TIME_ZONE: ${MONLET_WEB_TIME_ZONE:-UTC}
       MONLET_WEB_AUTH_USERNAME: ${MONLET_WEB_AUTH_USERNAME:-}
       MONLET_WEB_AUTH_PASSWORD: ${MONLET_WEB_AUTH_PASSWORD:-}
+      MONLET_WEB_SESSION_SECRET: ${MONLET_WEB_SESSION_SECRET:-}
+      MONLET_WEB_SESSION_TTL_SEC: ${MONLET_WEB_SESSION_TTL_SEC:-604800}
       MONLET_WEB_TRUST_PROXY_AUTH: ${MONLET_WEB_TRUST_PROXY_AUTH:-false}
       TZ: ${MONLET_WEB_TIME_ZONE:-UTC}
     ports:

docs/operations/deployment.md

@@ -50,7 +50,7 @@ Web is read-only for monitoring state in v1. Agent admission and blacklist contr
 
 Auth plan:
 
-- reverse proxy basic auth or equivalent gateway auth;
+- local web login or equivalent gateway auth through a trusted reverse proxy;
 - no full RBAC in v1.
 
 ## Observability

docs/ops/deployment-checklist.md

@@ -5,6 +5,7 @@ A pragmatic list for the first production-ish Monlet stand-up. Adjust per enviro
 ## 1. Secrets
 
 - [ ] Generate `MONLET_AUTH_TOKEN` (`openssl rand -hex 32`) and store it in your secret manager.
+- [ ] Generate `MONLET_WEB_SESSION_SECRET` (`openssl rand -hex 32`) if local web login is enabled.
 - [ ] Generate a PostgreSQL password and store it in your secret manager.
 - [ ] Decide which notifier channels you want and gather their credentials (Telegram bot token + chat id, webhook URL, Alertmanager URL).
 
@@ -41,7 +42,8 @@ A pragmatic list for the first production-ish Monlet stand-up. Adjust per enviro
 ## 6. Web UI
 
 - [ ] Set `MONLET_API_BASE_URL`, `MONLET_API_TOKEN`, and optional `MONLET_WEB_TIME_ZONE` for the web container.
-- [ ] Protect the web UI with either `MONLET_WEB_AUTH_USERNAME`/`MONLET_WEB_AUTH_PASSWORD` or a trusted reverse proxy that sets `X-Forwarded-User` plus `MONLET_WEB_TRUST_PROXY_AUTH=true`.
+- [ ] Protect the web UI with either local login (`MONLET_WEB_AUTH_USERNAME`, `MONLET_WEB_AUTH_PASSWORD`, `MONLET_WEB_SESSION_SECRET`) or a trusted reverse proxy that sets `X-Forwarded-User` plus `MONLET_WEB_TRUST_PROXY_AUTH=true`.
+- [ ] Rate-limit `/login` at the reverse proxy when local web login is enabled.
 - [ ] Set optional `MONLET_NOTIFICATION_TIME_ZONE` for human-readable notifier timestamps.
 - [ ] Browse `/`, `/agents`, `/checks`, `/incidents`, `/events`, `/outbox`.
 - [ ] Verify the web image does not log the token (`docker logs` greps clean).

server/monlet_server/api/events.py

@@ -7,8 +7,10 @@ from fastapi import APIRouter, Depends, HTTPException, Query, status
 from sqlalchemy import and_, or_, select
 from sqlalchemy.ext.asyncio import AsyncSession
 
+from .. import metrics
 from ..agent_auth import AgentCredential, get_agent_credential
 from ..db import get_session
+from ..logging_config import get_logger
 from ..models import Event
 from ..schemas import (
     ID_PATTERN,
@@ -22,6 +24,8 @@ from ..timeutil import to_utc
 
 router = APIRouter()
 
+log = get_logger("monlet.events")
+
 
 def _encode_cursor(received_at: datetime, event_id) -> str:
     raw = f"{to_utc(received_at).isoformat()}|{event_id}".encode()
@@ -52,12 +56,24 @@ async def events(
 ) -> EventBatchAcceptedResponse:
     agent = await accepted_agent_for_key(session, credential)
     if agent is None:
-        await session.commit()
+        raise HTTPException(status_code=403, detail="agent not accepted")
-        return EventBatchAcceptedResponse(accepted=0, deduplicated=0)
     accepted = 0
     deduplicated = 0
     for ev in body.events:
+        try:
+            async with session.begin_nested():
                 ok = await ingest_event(session, agent.agent_id, ev)
+        except HTTPException as exc:
+            if exc.status_code != 400:
+                raise
+            log.warning(
+                "event_rejected",
+                check_id=ev.check_id,
+                agent_id=agent.agent_id,
+                code=exc.detail,
+            )
+            metrics.events_total.labels(result="rejected").inc()
+            continue
         if ok:
             accepted += 1
         else:

server/monlet_server/errors.py

@@ -46,11 +46,12 @@ def install_error_handlers(app: FastAPI) -> None:
 
     @app.exception_handler(StarletteHTTPException)
     async def _http(request: Request, exc: StarletteHTTPException):
-        code = (
+        code: ErrorCode = _STATUS_TO_CODE.get(exc.status_code, "internal")
-            "agent_blocked"
+        if exc.status_code == 403:
-            if exc.status_code == 403 and exc.detail == "agent blocked"
+            if exc.detail == "agent blocked":
-            else _STATUS_TO_CODE.get(exc.status_code, "internal")
+                code = "agent_blocked"
-        )
+            elif exc.detail == "agent not accepted":
+                code = "agent_not_accepted"
         msg = exc.detail if isinstance(exc.detail, str) else code
         return error_response(request, exc.status_code, code, msg)
 

server/monlet_server/metrics.py

@@ -1,4 +1,4 @@
-from prometheus_client import CollectorRegistry, Counter, Gauge, Histogram, generate_latest
+from prometheus_client import CollectorRegistry, Counter, Gauge, generate_latest
 from prometheus_client.exposition import CONTENT_TYPE_LATEST
 
 REGISTRY = CollectorRegistry()
@@ -42,12 +42,6 @@ notifications_total = Counter(
     ["notifier", "result"],
     registry=REGISTRY,
 )
-request_duration = Histogram(
-    "monlet_server_request_duration_seconds",
-    "Request duration seconds.",
-    ["path"],
-    registry=REGISTRY,
-)
 
 
 def render() -> tuple[bytes, str]:

server/monlet_server/schemas.py

@@ -1,7 +1,6 @@
 import re
 from datetime import datetime
 from typing import Literal
-from uuid import UUID
 
 from pydantic import BaseModel, ConfigDict, Field, field_validator
 
@@ -39,6 +38,7 @@ ErrorCode = Literal[
     "internal",
     "not_ready",
     "agent_blocked",
+    "agent_not_accepted",
 ]
 
 
@@ -233,12 +233,6 @@ class OutboxPage(PageEnvelope):
     items: list[NotificationOutboxItem]
 
 
-class UUIDStr(BaseModel):
-    """Helper to validate generated UUIDs."""
-
-    value: UUID
-
-
 class SystemSummaryResponse(BaseModel):
     agents_online: int
     agents_offline: int

server/monlet_server/services/ingestion.py

@@ -53,13 +53,7 @@ async def _blacklist_row(session: AsyncSession, cred: AgentCredential) -> AgentB
     return res.scalar_one_or_none()
 
 
-async def reject_if_blocked(
+async def reject_if_blocked(session: AsyncSession, cred: AgentCredential) -> None:
-    session: AsyncSession,
-    cred: AgentCredential,
-    agent_id: str,
-    hostname: str,
-    observed: datetime | None = None,
-) -> None:
     row = await _blacklist_row(session, cred)
     if row is None:
         return
@@ -108,7 +102,7 @@ async def upsert_agent_heartbeat(
     session: AsyncSession, hb: HeartbeatRequest, cred: AgentCredential
 ) -> str:
     observed = to_utc(hb.observed_at)
-    await reject_if_blocked(session, cred, hb.agent_id, hb.hostname, observed)
+    await reject_if_blocked(session, cred)
 
     features = hb.features.model_dump()
     res = await session.execute(select(Agent).where(Agent.accepted_key_hash == cred.key_hash))
@@ -330,7 +324,6 @@ async def ingest_event(
             incident_key=incident_key,
         )
         await session.execute(stmt)
-        metrics.events_total.labels(result="accepted").inc()
     else:
         metrics.events_total.labels(result="unchanged").inc()
 
@@ -366,6 +359,9 @@ async def ingest_event(
         return False
 
     transition, inc = await _apply_incident(session, agent_id, ev, incident_key)
+    # Count accepted only after _apply_incident: an incident_key collision raises
+    # 400 and the savepoint rollback discards the Event insert above.
+    metrics.events_total.labels(result="accepted").inc()
     if transition and inc is not None and ev.notifications_enabled:
         event_type = "resolved" if ev.status == "ok" else "firing"
         await _enqueue_outbox(

server/monlet_server/services/notifier_worker.py

@@ -208,9 +208,16 @@ async def tick(
         await session.commit()
     if not rows:
         return 0
-    await asyncio.gather(
+
-        *(_process_row(sm, registry, r, settings.notifier_max_attempts) for r in rows)
+    # Bound concurrent DB sessions: each _process_row opens its own session, so an
-    )
+    # unbounded gather over a large batch can exhaust the connection pool.
+    sem = asyncio.Semaphore(settings.notifier_concurrency)
+
+    async def _guarded(row: dict) -> None:
+        async with sem:
+            await _process_row(sm, registry, row, settings.notifier_max_attempts)
+
+    await asyncio.gather(*(_guarded(r) for r in rows))
     return len(rows)
 
 

server/monlet_server/services/notifiers/http.py

@@ -1,7 +1,5 @@
 from __future__ import annotations
 
-import httpx
-
 
 def classify_response(status: int) -> tuple[bool, bool]:
     """Return (ok, permanent_failure)."""
@@ -10,9 +8,3 @@ def classify_response(status: int) -> tuple[bool, bool]:
     if 400 <= status < 500 and status not in (408, 425, 429):
         return False, True
     return False, False
-
-
-def classify_exception(exc: Exception) -> tuple[bool, bool]:
-    if isinstance(exc, httpx.TimeoutException | httpx.NetworkError):
-        return False, False
-    return False, False

server/monlet_server/settings.py

@@ -1,4 +1,4 @@
-from pydantic import field_validator
+from pydantic import Field, field_validator
 from pydantic_settings import BaseSettings, SettingsConfigDict
 
 from .timeutil import validate_time_zone
@@ -40,6 +40,7 @@ class Settings(BaseSettings):
     notifier_batch_size: int = 20
     notifier_max_attempts: int = 8
     notifier_lease_sec: int = 300
+    notifier_concurrency: int = Field(default=4, ge=1)
     notifier_http_timeout_sec: float = 10.0
     notification_time_zone: str = "UTC"
 

server/tests/test_agent_admission.py

@@ -22,14 +22,14 @@ async def test_unknown_heartbeat_creates_pending_agent(app_client, auth_headers,
 
 
 @pytest.mark.asyncio
-async def test_pending_events_are_dropped_until_accept(app_client, auth_headers, session):
+async def test_pending_events_are_rejected_until_accept(app_client, auth_headers, session):
     await app_client.post("/api/v1/heartbeat", json=make_heartbeat(), headers=auth_headers)
     ev = make_event(status="critical", exit_code=2)
     r = await app_client.post(
         "/api/v1/events", json={"agent_id": "agent-1", "events": [ev]}, headers=auth_headers
     )
-    assert r.status_code == 202
+    assert r.status_code == 403
-    assert r.json() == {"accepted": 0, "deduplicated": 0}
+    assert r.json()["error"]["code"] == "agent_not_accepted"
     n = (await session.execute(select(func.count()).select_from(Check))).scalar_one()
     assert n == 0
 

server/tests/test_events_ingestion.py

@@ -23,13 +23,13 @@ async def test_event_accepted_and_state(app_client, ui_auth_headers, session):
 
 
 @pytest.mark.asyncio
-async def test_event_before_heartbeat_is_dropped(app_client, auth_headers, session):
+async def test_event_before_heartbeat_is_rejected(app_client, auth_headers, session):
     ev = make_event(status="warning", exit_code=1)
     r = await app_client.post(
         "/api/v1/events", json={"agent_id": "agent-replay", "events": [ev]}, headers=auth_headers
     )
-    assert r.status_code == 202
+    assert r.status_code == 403
-    assert r.json() == {"accepted": 0, "deduplicated": 0}
+    assert r.json()["error"]["code"] == "agent_not_accepted"
     n = (await session.execute(select(func.count()).select_from(Agent))).scalar_one()
     assert n == 0
 

server/tests/test_incident_key.py

@@ -18,8 +18,8 @@ async def test_incident_key_ownership_rejected(app_client, ui_auth_headers):
     r2 = await app_client.post(
         "/api/v1/events", json={"agent_id": "agent-2", "events": [ev2]}, headers=agent_2_headers
     )
-    assert r2.status_code == 400
+    assert r2.status_code == 202
-    assert r2.json()["error"]["code"] == "validation"
+    assert r2.json() == {"accepted": 0, "deduplicated": 0}
 
 
 @pytest.mark.asyncio

web/README.md

@@ -36,11 +36,17 @@ Do not install Node packages globally. Project dependencies live in `web/node_mo
 | `MONLET_API_BASE_URL` | `http://127.0.0.1:8000` | Server base URL (server-side fetch) |
 | `MONLET_API_TOKEN` | _(none)_ | Bearer token sent to server |
 | `MONLET_WEB_TIME_ZONE` | `UTC` | IANA timezone used to render timestamps |
-| `MONLET_WEB_AUTH_USERNAME` | _(none)_ | Optional Basic Auth username for the web UI |
+| `MONLET_WEB_AUTH_USERNAME` | _(none)_ | Optional local web login username |
-| `MONLET_WEB_AUTH_PASSWORD` | _(none)_ | Optional Basic Auth password for the web UI |
+| `MONLET_WEB_AUTH_PASSWORD` | _(none)_ | Optional local web login password |
+| `MONLET_WEB_SESSION_SECRET` | _(none)_ | Secret used to sign the web session cookie; required with local web login; at least 32 bytes |
+| `MONLET_WEB_SESSION_TTL_SEC` | `604800` | Local web session lifetime |
 | `MONLET_WEB_TRUST_PROXY_AUTH` | `false` | Trust `X-Forwarded-User` from an upstream auth proxy |
 
-The API token never reaches the browser — all server calls happen in Server Components / route handlers. Agent admission actions are blocked unless Basic Auth is configured or trusted proxy auth is enabled.
+The API token never reaches the browser — all server calls happen in Server Components / route handlers. Agent admission actions are blocked unless local web login is configured or trusted proxy auth is enabled.
+
+Production deployments should rate-limit `/login` at the reverse proxy. The app applies constant-time credential checks and a small fixed delay on failed login attempts, but it is not a full auth gateway.
+
+The timezone switcher stores the selected mode in browser cookies, so each operator keeps their own default/browser/UTC preference.
 
 ## Pages
 

web/src/app/agents/[agent_id]/page.tsx

@@ -1,8 +1,10 @@
+import { redirect } from "next/navigation";
+
 import { AgentDetailBrowser } from "@/components/AgentDetailBrowser";
 import { ErrorPanel } from "@/components/DataPanel";
 import { ApiError, api } from "@/lib/api";
 import { loadChecks } from "@/lib/loaders";
-import { normalizeBack, normalizePageNumber, normalizePageSize } from "@/lib/pagination";
+import { INFINITE_BATCH_SIZE } from "@/lib/pagination";
 
 export const dynamic = "force-dynamic";
 
@@ -38,22 +40,19 @@ export default async function AgentDetailPage({
   const sp = await searchParams;
   const sort = normalizeSort(sp.sort);
   const tab = normalizeTab(sp.tab);
-  const eventsLimit = normalizePageSize(sp.events_limit, 50);
+  if (sp.cursor || sp.back || sp.page || sp.events_limit) {
-  const eventsCursor = sp.cursor;
+    redirect(agentDetailHref({ agentId: agent_id, tab, sort }));
-  const eventsPage = normalizePageNumber(sp.page);
+  }
-  const eventsBack = normalizeBack(sp.back);
 
   let agent;
   let checks;
-  let events: EventsPage = { items: [], next_cursor: null, prev_cursor: null };
+  let events: EventsPage = { items: [], next_cursor: null };
   try {
     [agent, checks] = await Promise.all([api.getAgent(agent_id), loadChecks(agent_id)]);
     if (tab === "events") {
       events = await api.queryEvents({
         agent_id,
-        limit: eventsLimit,
+        limit: INFINITE_BATCH_SIZE,
-        cursor: eventsCursor,
-        back: eventsBack,
       });
     }
   } catch (e) {
@@ -64,15 +63,24 @@ export default async function AgentDetailPage({
   }
   return (
     <AgentDetailBrowser
-      key={`${agent_id}:${tab}:${sort}:${eventsLimit}:${eventsCursor ?? ""}:${eventsBack ? "b" : "f"}:${eventsPage}`}
+      key={`${agent_id}:${tab}:${sort}`}
       initialData={{ agent, checks }}
       initialEvents={events}
       sort={sort}
       tab={tab}
-      eventsLimit={eventsLimit}
-      eventsCursor={eventsCursor}
-      eventsBack={eventsBack}
-      eventsPage={eventsPage}
     />
   );
 }
+
+function agentDetailHref({
+  agentId,
+  tab,
+  sort,
+}: {
+  agentId: string;
+  tab: TabKey;
+  sort: SortKey;
+}) {
+  const qs = new URLSearchParams({ tab, sort });
+  return `/agents/${encodeURIComponent(agentId)}?${qs.toString()}`;
+}

web/src/app/api/agents/action/route.ts

@@ -2,6 +2,7 @@ import { NextRequest, NextResponse } from "next/server";
 
 import { api } from "@/lib/api";
 import { jsonError } from "@/lib/poll-response";
+import { assertSameOrigin } from "@/lib/web-auth-server";
 
 export const dynamic = "force-dynamic";
 
@@ -16,6 +17,12 @@ type Action =
   | "blacklist_clear";
 
 export async function POST(request: NextRequest) {
+  if (!assertSameOrigin(request)) {
+    return NextResponse.json(
+      { error: { code: "forbidden", message: "cross-origin request rejected" } },
+      { status: 403 },
+    );
+  }
   try {
     const body = (await request.json()) as {
       action?: Action;

web/src/app/api/auth/login/route.ts

@@ -0,0 +1,58 @@
+import { NextRequest, NextResponse } from "next/server";
+
+import {
+  cookieSecure,
+  constantTimeEqualString,
+  createSessionToken,
+  delayLoginFailure,
+  localAuthEnabled,
+  proxyAuthEnabled,
+  sanitizeNext,
+  SESSION_COOKIE,
+  sessionTtlSec,
+} from "@/lib/web-auth";
+import { assertSameOrigin } from "@/lib/web-auth-server";
+
+export const dynamic = "force-dynamic";
+
+export async function POST(request: NextRequest) {
+  if (!assertSameOrigin(request)) return new NextResponse("forbidden", { status: 403 });
+  const form = await request.formData();
+  const next = sanitizeNext(stringField(form, "next"));
+  if (proxyAuthEnabled() || !localAuthEnabled()) return redirectLogin(request, next, "config");
+
+  const username = stringField(form, "username");
+  const password = stringField(form, "password");
+  const [usernameOk, passwordOk] = await Promise.all([
+    constantTimeEqualString(username, process.env.MONLET_WEB_AUTH_USERNAME ?? ""),
+    constantTimeEqualString(password, process.env.MONLET_WEB_AUTH_PASSWORD ?? ""),
+  ]);
+  if (!usernameOk || !passwordOk) {
+    await delayLoginFailure();
+    return redirectLogin(request, next, "invalid");
+  }
+
+  const ttl = sessionTtlSec();
+  const token = await createSessionToken(username, process.env.MONLET_WEB_SESSION_SECRET ?? "", ttl);
+  const response = NextResponse.redirect(new URL(next, request.url), { status: 303 });
+  response.cookies.set(SESSION_COOKIE, token, {
+    httpOnly: true,
+    sameSite: "lax",
+    secure: cookieSecure(request),
+    path: "/",
+    maxAge: ttl,
+  });
+  return response;
+}
+
+function redirectLogin(request: NextRequest, next: string, error: string) {
+  const url = new URL("/login", request.url);
+  url.searchParams.set("next", next);
+  url.searchParams.set("error", error);
+  return NextResponse.redirect(url, { status: 303 });
+}
+
+function stringField(form: FormData, name: string): string {
+  const value = form.get(name);
+  return typeof value === "string" ? value : "";
+}

web/src/app/api/auth/logout/route.ts

@@ -0,0 +1,19 @@
+import { NextRequest, NextResponse } from "next/server";
+
+import { cookieSecure, SESSION_COOKIE } from "@/lib/web-auth";
+import { assertSameOrigin } from "@/lib/web-auth-server";
+
+export const dynamic = "force-dynamic";
+
+export async function POST(request: NextRequest) {
+  if (!assertSameOrigin(request)) return new NextResponse("forbidden", { status: 403 });
+  const response = NextResponse.redirect(new URL("/login", request.url), { status: 303 });
+  response.cookies.set(SESSION_COOKIE, "", {
+    httpOnly: true,
+    sameSite: "lax",
+    secure: cookieSecure(request),
+    path: "/",
+    maxAge: 0,
+  });
+  return response;
+}

web/src/app/api/poll/events/route.ts

@@ -1,21 +1,20 @@
 import { NextResponse } from "next/server";
 
 import { loadEvents } from "@/lib/loaders";
-import { normalizePageSize } from "@/lib/pagination";
+import { normalizeFetchLimit } from "@/lib/pagination";
 import { jsonError } from "@/lib/poll-response";
 
 export const dynamic = "force-dynamic";
 
 export async function GET(request: Request) {
   const url = new URL(request.url);
-  const limit = normalizePageSize(url.searchParams.get("limit"), 50);
+  const limit = normalizeFetchLimit(url.searchParams.get("limit"));
 
   try {
     const data = await loadEvents({
       agentId: url.searchParams.get("agent_id") ?? undefined,
       checkId: url.searchParams.get("check_id") ?? undefined,
       cursor: url.searchParams.get("cursor") ?? undefined,
-      back: url.searchParams.get("back") === "1",
       limit,
     });
     return NextResponse.json(data);

web/src/app/api/poll/incidents/route.ts

@@ -15,7 +15,6 @@ export async function GET(request: Request) {
   const stateParam = url.searchParams.get("state");
   const state = stateParam === "open" || stateParam === "resolved" ? stateParam : undefined;
   const cursor = url.searchParams.get("cursor") ?? undefined;
-  const back = url.searchParams.get("back") === "1";
   const limitRaw = url.searchParams.get("limit");
   const parsed = limitRaw ? Number(limitRaw) : undefined;
   const limit = parsed && Number.isFinite(parsed) ? Math.min(Math.max(parsed, 1), 500) : 100;
@@ -28,7 +27,7 @@ export async function GET(request: Request) {
   const direction = dirRaw && DIRS.has(dirRaw) ? (dirRaw as "asc" | "desc") : undefined;
 
   try {
-    const page = await loadIncidentsPage({ state, cursor, back, limit, sort, direction });
+    const page = await loadIncidentsPage({ state, cursor, limit, sort, direction });
     return NextResponse.json(page);
   } catch (e) {
     return jsonError(e);

web/src/app/api/poll/outbox/route.ts

@@ -12,7 +12,6 @@ const DIRS = new Set(["asc", "desc"]);
 export async function GET(request: Request) {
   const url = new URL(request.url);
   const cursor = url.searchParams.get("cursor") ?? undefined;
-  const back = url.searchParams.get("back") === "1";
   const limitRaw = url.searchParams.get("limit");
   const parsed = limitRaw ? Number(limitRaw) : undefined;
   const limit = parsed && Number.isFinite(parsed) ? Math.min(Math.max(parsed, 1), 500) : 100;
@@ -26,7 +25,6 @@ export async function GET(request: Request) {
   try {
     const page = await loadOutboxPage({
       cursor,
-      back,
       limit,
       state: state as never,
       sort,

web/src/app/checks/[check_id]/page.tsx

@@ -1,8 +1,10 @@
+import { redirect } from "next/navigation";
+
 import { CheckDetailBrowser } from "@/components/CheckDetailBrowser";
 import { ErrorPanel } from "@/components/DataPanel";
 import { api } from "@/lib/api";
 import { loadInventory } from "@/lib/loaders";
-import { normalizeBack, normalizePageNumber, normalizePageSize } from "@/lib/pagination";
+import { INFINITE_BATCH_SIZE } from "@/lib/pagination";
 
 export const dynamic = "force-dynamic";
 
@@ -29,21 +31,18 @@ export default async function CheckDetailPage({
   const { check_id } = await params;
   const sp = await searchParams;
   const tab = normalizeTab(sp.tab);
-  const eventsLimit = normalizePageSize(sp.events_limit, 50);
+  if (sp.cursor || sp.back || sp.page || sp.events_limit) {
-  const eventsCursor = sp.cursor;
+    redirect(checkDetailHref({ checkId: check_id, tab }));
-  const eventsPage = normalizePageNumber(sp.page);
+  }
-  const eventsBack = normalizeBack(sp.back);
 
   let data;
-  let events: EventsPage = { items: [], next_cursor: null, prev_cursor: null };
+  let events: EventsPage = { items: [], next_cursor: null };
   try {
     data = await loadInventory();
     if (tab === "events") {
       events = await api.queryEvents({
         check_id,
-        limit: eventsLimit,
+        limit: INFINITE_BATCH_SIZE,
-        cursor: eventsCursor,
-        back: eventsBack,
       });
     }
   } catch (e) {
@@ -52,15 +51,22 @@ export default async function CheckDetailPage({
 
   return (
     <CheckDetailBrowser
-      key={`${check_id}:${tab}:${eventsLimit}:${eventsCursor ?? ""}:${eventsBack ? "b" : "f"}:${eventsPage}`}
+      key={`${check_id}:${tab}`}
       checkId={check_id}
       initialData={data}
       initialEvents={events}
       tab={tab}
-      eventsLimit={eventsLimit}
-      eventsCursor={eventsCursor}
-      eventsBack={eventsBack}
-      eventsPage={eventsPage}
     />
   );
 }
+
+function checkDetailHref({
+  checkId,
+  tab,
+}: {
+  checkId: string;
+  tab: TabKey;
+}) {
+  const qs = new URLSearchParams({ tab });
+  return `/checks/${encodeURIComponent(checkId)}?${qs.toString()}`;
+}

web/src/app/events/page.tsx

@@ -1,7 +1,9 @@
+import { redirect } from "next/navigation";
+
 import { ErrorPanel } from "@/components/DataPanel";
 import { EventsBrowser } from "@/components/EventsBrowser";
 import { loadEvents } from "@/lib/loaders";
-import { normalizeBack, normalizePageNumber, normalizePageSize } from "@/lib/pagination";
+import { INFINITE_BATCH_SIZE } from "@/lib/pagination";
 
 export const dynamic = "force-dynamic";
 
@@ -18,31 +20,39 @@ export default async function EventsPage({
   }>;
 }) {
   const sp = await searchParams;
-  const limit = normalizePageSize(sp.limit);
+  if (sp.cursor || sp.back || sp.page || sp.limit) {
-  const page = normalizePageNumber(sp.page);
+    redirect(eventsHref({ agentId: sp.agent_id, checkId: sp.check_id }));
-  const back = normalizeBack(sp.back);
+  }
   let data;
   try {
     data = await loadEvents({
       agentId: sp.agent_id,
       checkId: sp.check_id,
-      cursor: sp.cursor,
+      limit: INFINITE_BATCH_SIZE,
-      back,
-      limit,
     });
   } catch (e) {
     return <ErrorPanel error={e} />;
   }
   return (
     <EventsBrowser
-      key={`${sp.cursor ?? ""}:${back ? "b" : "f"}:${page}:${sp.agent_id ?? ""}:${sp.check_id ?? ""}:${limit}`}
+      key={`${sp.agent_id ?? ""}:${sp.check_id ?? ""}`}
       initialData={data}
-      cursor={sp.cursor}
-      back={back}
       agentId={sp.agent_id}
       checkId={sp.check_id}
-      limit={limit}
-      page={page}
     />
   );
 }
+
+function eventsHref({
+  agentId,
+  checkId,
+}: {
+  agentId?: string;
+  checkId?: string;
+}) {
+  const qs = new URLSearchParams();
+  if (agentId) qs.set("agent_id", agentId);
+  if (checkId) qs.set("check_id", checkId);
+  const query = qs.toString();
+  return query ? `/events?${query}` : "/events";
+}

web/src/app/incidents/page.tsx

@@ -1,7 +1,9 @@
+import { redirect } from "next/navigation";
+
 import { ErrorPanel } from "@/components/DataPanel";
 import { IncidentsBrowser } from "@/components/IncidentsBrowser";
 import { loadIncidentsPage, loadInventory } from "@/lib/loaders";
-import { normalizeBack, normalizePageNumber, normalizePageSize } from "@/lib/pagination";
+import { INFINITE_BATCH_SIZE } from "@/lib/pagination";
 
 export const dynamic = "force-dynamic";
 
@@ -25,14 +27,14 @@ export default async function IncidentsPage({
       ? sp.sort
       : "status";
   const direction = sp.direction === "asc" || sp.direction === "desc" ? sp.direction : "desc";
-  const limit = normalizePageSize(sp.limit);
+  if (sp.cursor || sp.back || sp.page || sp.limit) {
-  const pageNumber = normalizePageNumber(sp.page);
+    redirect(incidentsHref({ state, sort, direction }));
-  const back = normalizeBack(sp.back);
+  }
   let page;
   let inventory;
   try {
     [page, inventory] = await Promise.all([
-      loadIncidentsPage({ state, cursor: sp.cursor, back, limit, sort, direction }),
+      loadIncidentsPage({ state, limit: INFINITE_BATCH_SIZE, sort, direction }),
       loadInventory(),
     ]);
   } catch (e) {
@@ -40,16 +42,28 @@ export default async function IncidentsPage({
   }
   return (
     <IncidentsBrowser
-      key={`${sp.cursor ?? ""}:${back ? "b" : "f"}:${pageNumber}:${state ?? ""}:${sort}:${direction}:${limit}`}
+      key={`${state ?? ""}:${sort}:${direction}`}
       initialData={page}
       initialInventory={inventory}
       state={state}
-      cursor={sp.cursor}
-      back={back}
       sort={sort}
       direction={direction}
-      limit={limit}
-      page={pageNumber}
     />
   );
 }
+
+function incidentsHref({
+  state,
+  sort,
+  direction,
+}: {
+  state?: "open" | "resolved";
+  sort: "status" | "opened_at" | "resolved_at";
+  direction: "asc" | "desc";
+}) {
+  const qs = new URLSearchParams();
+  if (state) qs.set("state", state);
+  qs.set("sort", sort);
+  qs.set("direction", direction);
+  return `/incidents?${qs.toString()}`;
+}

web/src/app/layout.tsx

@@ -3,8 +3,9 @@ import type { ReactNode } from "react";
 import { TimeZoneProvider } from "@/components/TimeZoneControls";
 import { TopNav } from "@/components/TopNav";
 import { loadSystemSummary } from "@/lib/loaders";
-import { getUiTimeZone } from "@/lib/timezone";
+import { getInitialTimeZonePreference, getUiTimeZone } from "@/lib/timezone";
 import type { SystemSummaryData } from "@/lib/view-types";
+import { requestHasAppAccess, showLocalLogout } from "@/lib/web-auth-server";
 import "./globals.css";
 
 export const metadata: Metadata = {
@@ -24,13 +25,19 @@ async function safeLoadSystemSummary(): Promise<SystemSummaryData | null> {
 
 export default async function RootLayout({ children }: { children: ReactNode }) {
   const configuredTimeZone = getUiTimeZone();
-  const summary = await safeLoadSystemSummary();
+  const timezonePreference = await getInitialTimeZonePreference();
+  const hasAppAccess = await requestHasAppAccess();
+  const summary = hasAppAccess ? await safeLoadSystemSummary() : null;
 
   return (
     <html lang="en" className="h-full antialiased">
       <body className="min-h-full flex flex-col bg-neutral-950 text-neutral-100 font-mono">
-        <TimeZoneProvider configuredTimeZone={configuredTimeZone}>
+        <TimeZoneProvider
-          <TopNav initialSummary={summary} />
+          configuredTimeZone={configuredTimeZone}
+          initialMode={timezonePreference.mode}
+          initialBrowserTimeZone={timezonePreference.browserTimeZone}
+        >
+          {hasAppAccess ? <TopNav initialSummary={summary} showLogout={showLocalLogout()} /> : null}
           <main className="flex-1 px-4 py-6">{children}</main>
         </TimeZoneProvider>
       </body>

web/src/app/login/page.tsx

@@ -0,0 +1,96 @@
+import { redirect } from "next/navigation";
+
+import {
+  localAuthEnabled,
+  localAuthMisconfigured,
+  proxyAuthEnabled,
+  sanitizeNext,
+} from "@/lib/web-auth";
+import { requestHasAppAccess } from "@/lib/web-auth-server";
+
+export const dynamic = "force-dynamic";
+
+export default async function LoginPage({
+  searchParams,
+}: {
+  searchParams: Promise<{ next?: string; error?: string }>;
+}) {
+  const sp = await searchParams;
+  const next = sanitizeNext(sp.next);
+
+  if (await requestHasAppAccess()) redirect(next);
+
+  const proxyMode = proxyAuthEnabled();
+  const enabled = localAuthEnabled();
+  const misconfigured = localAuthMisconfigured();
+  const error =
+    sp.error === "invalid"
+      ? "Invalid username or password."
+      : sp.error === "config" || misconfigured
+        ? "Web authentication is not configured correctly."
+        : null;
+
+  return (
+    <div className="mx-auto mt-24 w-full max-w-sm">
+      <div className="mb-6 flex items-center gap-3">
+        <span className="relative h-9 w-9 overflow-hidden rounded-md border border-emerald-400/40 bg-neutral-900 shadow-[inset_0_0_0_1px_rgba(255,255,255,0.05)]">
+          <span className="absolute left-2 top-2 h-1.5 w-1.5 rounded-full bg-emerald-300" />
+          <span className="absolute bottom-2 left-2 h-2 w-0.5 rounded-full bg-cyan-300/80" />
+          <span className="absolute bottom-2 left-3.5 h-4 w-0.5 rounded-full bg-emerald-300" />
+          <span className="absolute bottom-2 left-5 h-2.5 w-0.5 rounded-full bg-sky-300/80" />
+        </span>
+        <h1 className="text-2xl font-semibold">
+          mon<span className="text-emerald-300">let</span>
+        </h1>
+      </div>
+
+      <h2 className="mb-4 text-lg font-semibold">Sign in</h2>
+      {error ? (
+        <div className="mb-4 border border-red-900 bg-red-950/30 px-3 py-2 text-sm text-red-200">
+          {error}
+        </div>
+      ) : null}
+      {misconfigured ? (
+        <div className="border border-red-900 bg-red-950/30 px-3 py-2 text-sm text-red-200">
+          Web authentication is misconfigured.
+        </div>
+      ) : proxyMode ? (
+        <div className="border border-amber-900 bg-amber-950/20 px-3 py-2 text-sm text-amber-200">
+          Proxy authentication is enabled. This page is only shown when the proxy did not pass
+          an authenticated user.
+        </div>
+      ) : enabled ? (
+        <form action="/api/auth/login" method="post" className="space-y-3">
+          <input type="hidden" name="next" value={next} />
+          <label className="block text-sm">
+            <span className="mb-1 block text-neutral-400">username</span>
+            <input
+              name="username"
+              autoComplete="username"
+              className="w-full border border-neutral-800 bg-neutral-950 px-3 py-2 text-neutral-100 outline-none focus:border-neutral-500"
+            />
+          </label>
+          <label className="block text-sm">
+            <span className="mb-1 block text-neutral-400">password</span>
+            <input
+              name="password"
+              type="password"
+              autoComplete="current-password"
+              className="w-full border border-neutral-800 bg-neutral-950 px-3 py-2 text-neutral-100 outline-none focus:border-neutral-500"
+            />
+          </label>
+          <button
+            type="submit"
+            className="w-full border border-emerald-900 bg-emerald-950/40 px-3 py-2 text-sm text-emerald-200 hover:border-emerald-600"
+          >
+            Sign in
+          </button>
+        </form>
+      ) : (
+        <div className="border border-neutral-800 bg-neutral-950 px-3 py-2 text-sm text-neutral-300">
+          Local web authentication is disabled.
+        </div>
+      )}
+    </div>
+  );
+}

web/src/app/outbox/page.tsx

@@ -1,7 +1,9 @@
+import { redirect } from "next/navigation";
+
 import { ErrorPanel } from "@/components/DataPanel";
 import { OutboxBrowser } from "@/components/OutboxBrowser";
 import { loadOutboxPage } from "@/lib/loaders";
-import { normalizeBack, normalizePageNumber, normalizePageSize } from "@/lib/pagination";
+import { INFINITE_BATCH_SIZE } from "@/lib/pagination";
 
 export const dynamic = "force-dynamic";
 
@@ -18,27 +20,34 @@ export default async function OutboxPage({
   }>;
 }) {
   const sp = await searchParams;
-  const limit = normalizePageSize(sp.limit);
   const sort = sp.sort === "created_at" || sp.sort === "updated_at" ? sp.sort : "updated_at";
   const direction = sp.direction === "asc" || sp.direction === "desc" ? sp.direction : "desc";
-  const pageNumber = normalizePageNumber(sp.page);
+  if (sp.cursor || sp.back || sp.page || sp.limit) redirect(outboxHref({ sort, direction }));
-  const back = normalizeBack(sp.back);
   let page;
   try {
-    page = await loadOutboxPage({ cursor: sp.cursor, back, limit, sort, direction });
+    page = await loadOutboxPage({ limit: INFINITE_BATCH_SIZE, sort, direction });
   } catch (e) {
     return <ErrorPanel error={e} />;
   }
   return (
     <OutboxBrowser
-      key={`${sp.cursor ?? ""}:${back ? "b" : "f"}:${pageNumber}:${sort}:${direction}:${limit}`}
+      key={`${sort}:${direction}`}
       initialData={page}
-      cursor={sp.cursor}
-      back={back}
       sort={sort}
       direction={direction}
-      limit={limit}
-      page={pageNumber}
     />
   );
 }
+
+function outboxHref({
+  sort,
+  direction,
+}: {
+  sort: "created_at" | "updated_at";
+  direction: "asc" | "desc";
+}) {
+  const qs = new URLSearchParams();
+  qs.set("sort", sort);
+  qs.set("direction", direction);
+  return `/outbox?${qs.toString()}`;
+}

web/src/components/AgentDetailBrowser.tsx

@@ -1,33 +1,25 @@
 "use client";
 
 import Link from "next/link";
-import { useRouter } from "next/navigation";
+import { useCallback, useMemo } from "react";
-import { useMemo, useState } from "react";
 
 import { AgentFeatures } from "@/components/AgentFeatures";
 import { CheckEventsLink } from "@/components/CheckEventsLink";
-import { ClientPager } from "@/components/ClientPager";
 import { EmptyPanel, SummaryCell, Td, Th } from "@/components/DataPanel";
 import { DetailTabs } from "@/components/DetailTabs";
+import { InfiniteListFooter } from "@/components/InfiniteListFooter";
 import { LabelChips } from "@/components/Labels";
-import { PageSizeSelect } from "@/components/PageSizeSelect";
-import { Pager } from "@/components/Pager";
 import { DateTime } from "@/components/TimeZoneControls";
 import type { components } from "@/lib/api-types";
 import { statusBadgeClass } from "@/lib/format";
-import {
+import { INFINITE_BATCH_SIZE } from "@/lib/pagination";
-  DEFAULT_PAGE_SIZE,
+import { useInfiniteClientList, useInfiniteCursorList } from "@/lib/useInfiniteList";
-  clampPage,
-  pageCountOf,
-  paginate,
-  type PageSize,
-} from "@/lib/pagination";
 import { usePolling } from "@/lib/usePolling";
 
 type Agent = components["schemas"]["Agent"];
 type CheckState = components["schemas"]["CheckState"];
 type StoredEvent = components["schemas"]["StoredCheckResultEvent"];
-type EventsPageData = { items: StoredEvent[]; next_cursor?: string | null; prev_cursor?: string | null };
+type EventsPageData = { items: StoredEvent[]; next_cursor?: string | null };
 type SortKey = "severity" | "check_id" | "last_seen";
 type TabKey = "checks" | "events";
 type AgentDetailData = { agent: Agent; checks: CheckState[] };
@@ -45,19 +37,11 @@ export function AgentDetailBrowser({
   initialEvents,
   sort,
   tab,
-  eventsLimit,
-  eventsCursor,
-  eventsBack,
-  eventsPage,
 }: {
   initialData: AgentDetailData;
   initialEvents: EventsPageData;
   sort: SortKey;
   tab: TabKey;
-  eventsLimit: PageSize;
-  eventsCursor?: string;
-  eventsBack: boolean;
-  eventsPage: number;
 }) {
   const agentId = initialData.agent.agent_id;
   const pollUrl = `/api/poll/agent?agent_id=${encodeURIComponent(agentId)}`;
@@ -102,22 +86,14 @@ export function AgentDetailBrowser({
           {
             key: "events",
             label: "Events",
-            href: `/agents/${encodeURIComponent(agentId)}?tab=events&sort=${sort}&events_limit=${eventsLimit}`,
+            href: `/agents/${encodeURIComponent(agentId)}?tab=events&sort=${sort}`,
           },
         ]}
       />
       {tab === "checks" ? (
         <ChecksTable agentId={agentId} checks={checks} sort={sort} />
       ) : (
-        <EventsTable
+        <EventsTable agentId={agentId} initialEvents={initialEvents} />
-          agentId={agentId}
-          sort={sort}
-          eventsLimit={eventsLimit}
-          eventsCursor={eventsCursor}
-          eventsBack={eventsBack}
-          eventsPage={eventsPage}
-          initialEvents={initialEvents}
-        />
       )}
     </div>
   );
@@ -132,22 +108,19 @@ function ChecksTable({
   checks: CheckState[];
   sort: SortKey;
 }) {
-  const [pageSize, setPageSize] = useState<PageSize>(DEFAULT_PAGE_SIZE);
+  const {
-  const [page, setPage] = useState(1);
+    visibleItems: visible,
-  const onPageSizeChange = (next: PageSize) => {
+    hasMore,
-    setPageSize(next);
+    sentinelRef,
-    setPage(1);
+  } = useInfiniteClientList({
-  };
+    items: checks,
-  const pageCount = pageCountOf(checks.length, pageSize);
+    batchSize: INFINITE_BATCH_SIZE,
-  const currentPage = clampPage(page, pageCount);
+    resetKey: `${agentId}:${sort}`,
-  const visible = paginate(checks, currentPage, pageSize);
+  });
 
   if (checks.length === 0) return <EmptyPanel label="No checks" />;
   return (
     <section>
-      <div className="mb-3 flex justify-end">
-        <PageSizeSelect value={pageSize} onChange={onPageSizeChange} />
-      </div>
       <table className="w-full text-sm">
         <thead>
           <tr>
@@ -184,48 +157,45 @@ function ChecksTable({
           ))}
         </tbody>
       </table>
-      <ClientPager page={currentPage} pageCount={pageCount} onPageChange={setPage} />
+      <InfiniteListFooter sentinelRef={sentinelRef} hasMore={hasMore} />
     </section>
   );
 }
 
 function EventsTable({
   agentId,
-  sort,
-  eventsLimit,
-  eventsCursor,
-  eventsBack,
-  eventsPage,
   initialEvents,
 }: {
   agentId: string;
-  sort: SortKey;
-  eventsLimit: PageSize;
-  eventsCursor?: string;
-  eventsBack: boolean;
-  eventsPage: number;
   initialEvents: EventsPageData;
 }) {
-  const router = useRouter();
   const pollUrl = useMemo(() => {
-    const qs = new URLSearchParams({ agent_id: agentId, limit: String(eventsLimit) });
+    const qs = new URLSearchParams({ agent_id: agentId, limit: String(INFINITE_BATCH_SIZE) });
-    if (eventsCursor) qs.set("cursor", eventsCursor);
-    if (eventsBack) qs.set("back", "1");
     return `/api/poll/events?${qs.toString()}`;
-  }, [agentId, eventsCursor, eventsBack, eventsLimit]);
+  }, [agentId]);
-  const { data } = usePolling(initialEvents, pollUrl);
+  const getPageUrl = useCallback(
-
+    (cursor: string) => `${pollUrl}&cursor=${encodeURIComponent(cursor)}`,
-  const onLimitChange = (next: PageSize) => {
+    [pollUrl],
-    const qs = new URLSearchParams({ tab: "events", sort, events_limit: String(next) });
+  );
-    router.push(`/agents/${encodeURIComponent(agentId)}?${qs.toString()}`);
+  const {
-  };
+    items,
+    hasMore,
+    loadingMore,
+    error,
+    sentinelRef,
+    hasNewItems,
+    showLatest,
+  } = useInfiniteCursorList<StoredEvent>({
+    initialData: initialEvents,
+    firstPageUrl: pollUrl,
+    getPageUrl,
+    getKey: eventKey,
+    resetKey: agentId,
+  });
 
   return (
     <section>
-      <div className="mb-3 flex justify-end">
+      {items.length === 0 ? (
-        <PageSizeSelect value={eventsLimit} onChange={onLimitChange} />
-      </div>
-      {data.items.length === 0 ? (
         <EmptyPanel label="No events" />
       ) : (
         <table className="w-full text-sm">
@@ -239,7 +209,7 @@ function EventsTable({
             </tr>
           </thead>
           <tbody>
-            {data.items.map((e) => (
+            {items.map((e) => (
               <tr key={e.event_id}>
                 <Td>
                   <DateTime iso={e.observed_at} />
@@ -255,12 +225,13 @@ function EventsTable({
           </tbody>
         </table>
       )}
-      <Pager
+      <InfiniteListFooter
-        basePath={`/agents/${encodeURIComponent(agentId)}`}
+        sentinelRef={sentinelRef}
-        prevCursor={data.prev_cursor}
+        hasMore={hasMore}
-        nextCursor={data.next_cursor}
+        loading={loadingMore}
-        page={eventsPage}
+        error={error}
-        extra={{ tab: "events", sort, events_limit: String(eventsLimit) }}
+        hasNewItems={hasNewItems}
+        onShowLatest={showLatest}
       />
     </section>
   );
@@ -294,3 +265,7 @@ function sortChecks(checks: CheckState[], sort: SortKey): CheckState[] {
     return severityRank[a.status] - severityRank[b.status] || a.check_id.localeCompare(b.check_id);
   });
 }
+
+function eventKey(event: StoredEvent): string {
+  return event.event_id;
+}

web/src/components/AgentsBrowser.tsx

@@ -5,21 +5,16 @@ import { useMemo, useState } from "react";
 
 import { AgentFeatures } from "@/components/AgentFeatures";
 import { AgentLink } from "@/components/AgentLink";
-import { ClientPager } from "@/components/ClientPager";
 import { EmptyPanel, Td, Th } from "@/components/DataPanel";
+import { InfiniteListFooter } from "@/components/InfiniteListFooter";
 import { LabelChips } from "@/components/Labels";
 import { ListToolbar } from "@/components/ListToolbar";
 import { SortHeaderButton, type SortDir } from "@/components/SortHeaderButton";
 import { DateTime } from "@/components/TimeZoneControls";
 import type { components } from "@/lib/api-types";
 import { statusBadgeClass } from "@/lib/format";
-import {
+import { INFINITE_BATCH_SIZE } from "@/lib/pagination";
-  DEFAULT_PAGE_SIZE,
+import { useInfiniteClientList } from "@/lib/useInfiniteList";
-  clampPage,
-  pageCountOf,
-  paginate,
-  type PageSize,
-} from "@/lib/pagination";
 import { usePolling } from "@/lib/usePolling";
 import type { InventoryData } from "@/lib/view-types";
 
@@ -56,19 +51,18 @@ export function AgentsBrowser({ initialData }: { initialData: InventoryData }) {
     ({ agent }) => agent.rotation_pending,
   ).length;
 
-  const [pageSize, setPageSize] = useState<PageSize>(DEFAULT_PAGE_SIZE);
+  const {
-  const [page, setPage] = useState(1);
+    visibleItems: visible,
-  const pageCount = pageCountOf(rows.length, pageSize);
+    hasMore,
-  const currentPage = clampPage(page, pageCount);
+    sentinelRef,
-  const visible = paginate(rows, currentPage, pageSize);
+  } = useInfiniteClientList({
+    items: rows,
+    batchSize: INFINITE_BATCH_SIZE,
+    resetKey: `${query}:${sort}:${dir}`,
+  });
 
   const onQueryChange = (value: string) => {
     setQuery(value);
-    setPage(1);
-  };
-  const onPageSizeChange = (next: PageSize) => {
-    setPageSize(next);
-    setPage(1);
   };
   const chooseSort = (field: SortField) => {
     if (field === sort) {
@@ -77,7 +71,6 @@ export function AgentsBrowser({ initialData }: { initialData: InventoryData }) {
       setSort(field);
       setDir(fieldDefaultDir(field));
     }
-    setPage(1);
   };
   const runAction = async (
     action: string,
@@ -123,8 +116,6 @@ export function AgentsBrowser({ initialData }: { initialData: InventoryData }) {
         query={query}
         onQueryChange={onQueryChange}
         placeholder="search agent"
-        pageSize={pageSize}
-        onPageSizeChange={onPageSizeChange}
       />
       {rows.length === 0 ? (
         <EmptyPanel />
@@ -140,7 +131,7 @@ export function AgentsBrowser({ initialData }: { initialData: InventoryData }) {
           rotationPendingCount={allRotationPendingCount}
         />
       )}
-      <ClientPager page={currentPage} pageCount={pageCount} onPageChange={setPage} />
+      <InfiniteListFooter sentinelRef={sentinelRef} hasMore={hasMore} />
     </>
   );
 }
@@ -173,6 +164,8 @@ function AgentsTable({
     rotationPendingCount > 0
       ? `ROTATION: existing accepted keys will be replaced for ${rotationPendingCount} agents. Accept all pending agents including rotations?`
       : undefined;
+  const visiblePendingRows = rows.filter(({ agent }) => agent.admission_state === "pending");
+  const visibleAcceptedRows = rows.filter(({ agent }) => agent.admission_state === "accepted");
   return (
     <table className="w-full text-sm">
       <thead>
@@ -198,37 +191,76 @@ function AgentsTable({
         </tr>
       </thead>
       <tbody>
-        {pendingCount > 0 ? (
+        {visiblePendingRows.length > 0 ? (
           <tr>
             <td colSpan={6} className="border-b border-neutral-900 bg-amber-950/20 px-2 py-2 text-amber-300">
-              <div className="flex flex-wrap items-center gap-2">
+              <div className="flex flex-wrap items-center gap-3">
                 <span>{pendingCount} pending</span>
-                <ActionButton
+                {rotationPendingCount > 0 ? (
-                  label="✅"
+                  <span className="text-amber-500">{rotationPendingCount} replacements</span>
-                  title="Accept all"
+                ) : null}
+              </div>
+            </td>
+          </tr>
+        ) : null}
+        {visiblePendingRows.map((row) => (
+          <AgentTableRow key={row.agent.agent_id} row={row} busy={busy} runAction={runAction} />
+        ))}
+        {visiblePendingRows.length > 0 ? (
+          <tr>
+            <td colSpan={6} className="border-b border-neutral-900 bg-amber-950/10 px-2 py-3">
+              <div className="flex flex-wrap items-center gap-2">
+                <BulkActionButton
+                  label="✅ Accept all"
+                  title="Accept all pending agents"
                   disabled={busy !== null}
                   onClick={() =>
                     runAction("accept_all", undefined, acceptAllText, rotationPendingCount > 0)
                   }
                 />
-                <ActionButton
+                <BulkActionButton
-                  label="⛔"
+                  label="⛔ Block all"
-                  title="Block all"
+                  title="Block all pending agents"
                   disabled={busy !== null}
                   onClick={() => runAction("block_all", undefined, "Block all pending agents?")}
                 />
-                <ActionButton
+                <BulkActionButton
-                  label="🗑"
+                  label="🗑 Remove all"
-                  title="Remove all"
+                  title="Remove all pending agents"
                   disabled={busy !== null}
-                  onClick={() => runAction("remove_all", undefined, "Remove all pending agents and their data?")}
+                  onClick={() =>
+                    runAction("remove_all", undefined, "Remove all pending agents and their data?")
+                  }
                 />
               </div>
             </td>
           </tr>
         ) : null}
-        {rows.map(({ agent: a, checks }) => (
+        {visibleAcceptedRows.map((row) => (
-          <tr key={a.agent_id} className={a.admission_state === "pending" ? "bg-amber-950/10" : ""}>
+          <AgentTableRow key={row.agent.agent_id} row={row} busy={busy} runAction={runAction} />
+        ))}
+      </tbody>
+    </table>
+  );
+}
+
+function AgentTableRow({
+  row,
+  busy,
+  runAction,
+}: {
+  row: AgentRow;
+  busy: string | null;
+  runAction: (
+    action: string,
+    agentId?: string,
+    confirmText?: string,
+    includeRotation?: boolean,
+  ) => void;
+}) {
+  const { agent: a, checks } = row;
+  return (
+    <tr className={a.admission_state === "pending" ? "bg-amber-950/10" : ""}>
       <Td>
         <div className="flex items-center gap-2">
           <AgentLink agent={a} agentId={a.agent_id} />
@@ -255,9 +287,6 @@ function AgentsTable({
         <AgentActions agent={a} busy={busy} runAction={runAction} />
       </Td>
     </tr>
-        ))}
-      </tbody>
-    </table>
   );
 }
 
@@ -313,6 +342,31 @@ function AgentActions({
   );
 }
 
+function BulkActionButton({
+  label,
+  title,
+  disabled,
+  onClick,
+}: {
+  label: string;
+  title: string;
+  disabled: boolean;
+  onClick: () => void;
+}) {
+  return (
+    <button
+      type="button"
+      title={title}
+      aria-label={title}
+      disabled={disabled}
+      onClick={onClick}
+      className="rounded border border-neutral-700 bg-neutral-950 px-3 py-1.5 text-sm text-neutral-100 hover:border-neutral-400 disabled:opacity-40"
+    >
+      {label}
+    </button>
+  );
+}
+
 function ActionButton({
   label,
   title,

web/src/components/CheckDetailBrowser.tsx

@@ -1,32 +1,24 @@
 "use client";
 
 import Link from "next/link";
-import { useRouter } from "next/navigation";
+import { useCallback, useMemo } from "react";
-import { useMemo, useState } from "react";
 
 import { AgentLink } from "@/components/AgentLink";
 import { CheckStatusSummary } from "@/components/CheckStatusSummary";
-import { ClientPager } from "@/components/ClientPager";
 import { EmptyPanel, SummaryCell, Td, Th } from "@/components/DataPanel";
 import { DetailTabs } from "@/components/DetailTabs";
-import { PageSizeSelect } from "@/components/PageSizeSelect";
+import { InfiniteListFooter } from "@/components/InfiniteListFooter";
-import { Pager } from "@/components/Pager";
 import { DateTime } from "@/components/TimeZoneControls";
 import type { components } from "@/lib/api-types";
 import { groupChecks, statusSeverity, type Agent, type CheckGroup } from "@/lib/check-groups";
 import { statusBadgeClass } from "@/lib/format";
-import {
+import { INFINITE_BATCH_SIZE } from "@/lib/pagination";
-  DEFAULT_PAGE_SIZE,
+import { useInfiniteClientList, useInfiniteCursorList } from "@/lib/useInfiniteList";
-  clampPage,
-  pageCountOf,
-  paginate,
-  type PageSize,
-} from "@/lib/pagination";
 import { usePolling } from "@/lib/usePolling";
 import type { InventoryData } from "@/lib/view-types";
 
 type StoredEvent = components["schemas"]["StoredCheckResultEvent"];
-type EventsPageData = { items: StoredEvent[]; next_cursor?: string | null; prev_cursor?: string | null };
+type EventsPageData = { items: StoredEvent[]; next_cursor?: string | null };
 type TabKey = "agents" | "events";
 
 export function CheckDetailBrowser({
@@ -34,19 +26,11 @@ export function CheckDetailBrowser({
   initialData,
   initialEvents,
   tab,
-  eventsLimit,
-  eventsCursor,
-  eventsBack,
-  eventsPage,
 }: {
   checkId: string;
   initialData: InventoryData;
   initialEvents: EventsPageData;
   tab: TabKey;
-  eventsLimit: PageSize;
-  eventsCursor?: string;
-  eventsBack: boolean;
-  eventsPage: number;
 }) {
   const { data } = usePolling(initialData, "/api/poll/inventory");
   const agentsById = useMemo(
@@ -101,7 +85,7 @@ export function CheckDetailBrowser({
           {
             key: "events",
             label: "Events",
-            href: `/checks/${encodeURIComponent(checkId)}?tab=events&events_limit=${eventsLimit}`,
+            href: `/checks/${encodeURIComponent(checkId)}?tab=events`,
           },
         ]}
       />
@@ -112,10 +96,6 @@ export function CheckDetailBrowser({
           checkId={checkId}
           initialData={initialEvents}
           agentsById={agentsById}
-          eventsLimit={eventsLimit}
-          eventsCursor={eventsCursor}
-          eventsBack={eventsBack}
-          eventsPage={eventsPage}
         />
       )}
     </div>
@@ -133,21 +113,18 @@ function AgentsForCheck({ group }: { group: CheckGroup }) {
     [group.checks],
   );
 
-  const [pageSize, setPageSize] = useState<PageSize>(DEFAULT_PAGE_SIZE);
+  const {
-  const [page, setPage] = useState(1);
+    visibleItems: visible,
-  const onPageSizeChange = (next: PageSize) => {
+    hasMore,
-    setPageSize(next);
+    sentinelRef,
-    setPage(1);
+  } = useInfiniteClientList({
-  };
+    items: rows,
-  const pageCount = pageCountOf(rows.length, pageSize);
+    batchSize: INFINITE_BATCH_SIZE,
-  const currentPage = clampPage(page, pageCount);
+    resetKey: group.checkId,
-  const visible = paginate(rows, currentPage, pageSize);
+  });
 
   return (
     <section>
-      <div className="mb-3 flex justify-end">
-        <PageSizeSelect value={pageSize} onChange={onPageSizeChange} />
-      </div>
       <table className="w-full text-sm">
         <thead>
           <tr>
@@ -172,7 +149,7 @@ function AgentsForCheck({ group }: { group: CheckGroup }) {
           ))}
         </tbody>
       </table>
-      <ClientPager page={currentPage} pageCount={pageCount} onPageChange={setPage} />
+      <InfiniteListFooter sentinelRef={sentinelRef} hasMore={hasMore} />
     </section>
   );
 }
@@ -181,43 +158,41 @@ function EventsForCheck({
   checkId,
   initialData,
   agentsById,
-  eventsLimit,
-  eventsCursor,
-  eventsBack,
-  eventsPage,
 }: {
   checkId: string;
   initialData: EventsPageData;
   agentsById: Map<string, Agent>;
-  eventsLimit: PageSize;
-  eventsCursor?: string;
-  eventsBack: boolean;
-  eventsPage: number;
 }) {
-  const router = useRouter();
   const pollUrl = useMemo(() => {
     const qs = new URLSearchParams({
       check_id: checkId,
-      limit: String(eventsLimit),
+      limit: String(INFINITE_BATCH_SIZE),
     });
-    if (eventsCursor) qs.set("cursor", eventsCursor);
-    if (eventsBack) qs.set("back", "1");
     return `/api/poll/events?${qs.toString()}`;
-  }, [checkId, eventsCursor, eventsBack, eventsLimit]);
+  }, [checkId]);
-  const { data } = usePolling(initialData, pollUrl);
+  const getPageUrl = useCallback(
-
+    (cursor: string) => `${pollUrl}&cursor=${encodeURIComponent(cursor)}`,
-  const onLimitChange = (next: PageSize) => {
+    [pollUrl],
-    router.push(
-      `/checks/${encodeURIComponent(checkId)}?tab=events&events_limit=${next}`,
   );
-  };
+  const {
+    items,
+    hasMore,
+    loadingMore,
+    error,
+    sentinelRef,
+    hasNewItems,
+    showLatest,
+  } = useInfiniteCursorList<StoredEvent>({
+    initialData,
+    firstPageUrl: pollUrl,
+    getPageUrl,
+    getKey: eventKey,
+    resetKey: checkId,
+  });
 
   return (
     <section>
-      <div className="mb-3 flex justify-end">
+      {items.length === 0 ? (
-        <PageSizeSelect value={eventsLimit} onChange={onLimitChange} />
-      </div>
-      {data.items.length === 0 ? (
         <EmptyPanel label="No events" />
       ) : (
         <table className="w-full text-sm">
@@ -232,7 +207,7 @@ function EventsForCheck({
             </tr>
           </thead>
           <tbody>
-            {data.items.map((event) => {
+            {items.map((event) => {
               const agent = agentsById.get(event.agent_id);
               return (
                 <tr key={event.event_id}>
@@ -254,12 +229,13 @@ function EventsForCheck({
           </tbody>
         </table>
       )}
-      <Pager
+      <InfiniteListFooter
-        basePath={`/checks/${encodeURIComponent(checkId)}`}
+        sentinelRef={sentinelRef}
-        prevCursor={data.prev_cursor}
+        hasMore={hasMore}
-        nextCursor={data.next_cursor}
+        loading={loadingMore}
-        page={eventsPage}
+        error={error}
-        extra={{ tab: "events", events_limit: String(eventsLimit) }}
+        hasNewItems={hasNewItems}
+        onShowLatest={showLatest}
       />
     </section>
   );
@@ -268,3 +244,7 @@ function EventsForCheck({
 function hostName(row: { agent?: Agent; check: { agent_id: string } }): string {
   return row.agent?.hostname || row.check.agent_id;
 }
+
+function eventKey(event: StoredEvent): string {
+  return event.event_id;
+}

web/src/components/ChecksBrowser.tsx

@@ -4,8 +4,8 @@ import Link from "next/link";
 import { useMemo, useState } from "react";
 
 import { CheckStatusSummary } from "@/components/CheckStatusSummary";
-import { ClientPager } from "@/components/ClientPager";
 import { EmptyPanel, Td, Th } from "@/components/DataPanel";
+import { InfiniteListFooter } from "@/components/InfiniteListFooter";
 import { ListToolbar } from "@/components/ListToolbar";
 import { SortHeaderButton, type SortDir } from "@/components/SortHeaderButton";
 import { DateTime } from "@/components/TimeZoneControls";
@@ -14,13 +14,8 @@ import {
   groupChecks,
   type CheckGroup,
 } from "@/lib/check-groups";
-import {
+import { INFINITE_BATCH_SIZE } from "@/lib/pagination";
-  DEFAULT_PAGE_SIZE,
+import { useInfiniteClientList } from "@/lib/useInfiniteList";
-  clampPage,
-  pageCountOf,
-  paginate,
-  type PageSize,
-} from "@/lib/pagination";
 import { usePolling } from "@/lib/usePolling";
 import type { InventoryData } from "@/lib/view-types";
 
@@ -39,19 +34,18 @@ export function ChecksBrowser({ initialData }: { initialData: InventoryData }) {
     [dir, groups, query, sort],
   );
 
-  const [pageSize, setPageSize] = useState<PageSize>(DEFAULT_PAGE_SIZE);
+  const {
-  const [page, setPage] = useState(1);
+    visibleItems: visible,
-  const pageCount = pageCountOf(rows.length, pageSize);
+    hasMore,
-  const currentPage = clampPage(page, pageCount);
+    sentinelRef,
-  const visible = paginate(rows, currentPage, pageSize);
+  } = useInfiniteClientList({
+    items: rows,
+    batchSize: INFINITE_BATCH_SIZE,
+    resetKey: `${query}:${sort}:${dir}`,
+  });
 
   const onQueryChange = (value: string) => {
     setQuery(value);
-    setPage(1);
-  };
-  const onPageSizeChange = (next: PageSize) => {
-    setPageSize(next);
-    setPage(1);
   };
   const chooseSort = (field: SortField) => {
     if (field === sort) {
@@ -60,7 +54,6 @@ export function ChecksBrowser({ initialData }: { initialData: InventoryData }) {
       setSort(field);
       setDir(fieldDefaultDir(field));
     }
-    setPage(1);
   };
 
   return (
@@ -75,8 +68,6 @@ export function ChecksBrowser({ initialData }: { initialData: InventoryData }) {
         query={query}
         onQueryChange={onQueryChange}
         placeholder="search check"
-        pageSize={pageSize}
-        onPageSizeChange={onPageSizeChange}
       />
       {rows.length === 0 ? (
         <EmptyPanel />
@@ -125,7 +116,7 @@ export function ChecksBrowser({ initialData }: { initialData: InventoryData }) {
           </tbody>
         </table>
       )}
-      <ClientPager page={currentPage} pageCount={pageCount} onPageChange={setPage} />
+      <InfiniteListFooter sentinelRef={sentinelRef} hasMore={hasMore} />
     </>
   );
 }

web/src/components/ClientPager.tsx

@@ -1,39 +0,0 @@
-"use client";
-
-export function ClientPager({
-  page,
-  pageCount,
-  onPageChange,
-}: {
-  page: number;
-  pageCount: number;
-  onPageChange: (page: number) => void;
-}) {
-  const prev = () => onPageChange(Math.max(1, page - 1));
-  const next = () => onPageChange(Math.min(pageCount, page + 1));
-  const disabledPrev = page <= 1;
-  const disabledNext = page >= pageCount;
-  return (
-    <div className="mt-4 flex items-center justify-between text-sm">
-      <button
-        type="button"
-        onClick={prev}
-        disabled={disabledPrev}
-        className="text-blue-300 hover:text-blue-200 disabled:cursor-default disabled:text-neutral-600"
-      >
-        ← prev
-      </button>
-      <span className="text-neutral-500">
-        page {page} of {pageCount}
-      </span>
-      <button
-        type="button"
-        onClick={next}
-        disabled={disabledNext}
-        className="text-blue-300 hover:text-blue-200 disabled:cursor-default disabled:text-neutral-600"
-      >
-        next →
-      </button>
-    </div>
-  );
-}

web/src/components/EventsBrowser.tsx

@@ -1,58 +1,60 @@
 "use client";
 
 import Link from "next/link";
-import { useRouter } from "next/navigation";
+import { useCallback, useMemo } from "react";
-import { useMemo } from "react";
 
 import { CheckEventsLink } from "@/components/CheckEventsLink";
 import { EmptyPanel, PageHeader, SummaryCell, Td, Th } from "@/components/DataPanel";
-import { PageSizeSelect } from "@/components/PageSizeSelect";
+import { InfiniteListFooter } from "@/components/InfiniteListFooter";
-import { Pager } from "@/components/Pager";
 import { DateTime } from "@/components/TimeZoneControls";
 import type { components } from "@/lib/api-types";
 import { statusBadgeClass } from "@/lib/format";
-import { type PageSize } from "@/lib/pagination";
+import { INFINITE_BATCH_SIZE } from "@/lib/pagination";
-import { usePolling } from "@/lib/usePolling";
+import { useInfiniteCursorList } from "@/lib/useInfiniteList";
 
 type StoredEvent = components["schemas"]["StoredCheckResultEvent"];
-type EventsPageData = { items: StoredEvent[]; next_cursor?: string | null; prev_cursor?: string | null };
+type EventsPageData = { items: StoredEvent[]; next_cursor?: string | null };
 
 export function EventsBrowser({
   initialData,
-  cursor,
-  back,
   agentId,
   checkId,
-  limit,
-  page,
 }: {
   initialData: EventsPageData;
-  cursor?: string;
-  back: boolean;
   agentId?: string;
   checkId?: string;
-  limit: PageSize;
-  page: number;
 }) {
-  const router = useRouter();
+  const firstPageUrl = useMemo(
-  const url = useMemo(
+    () => eventsPollUrl({ agentId, checkId }),
-    () => eventsPollUrl({ cursor, back, agentId, checkId, limit }),
+    [agentId, checkId],
-    [agentId, back, checkId, cursor, limit],
   );
-  const { data } = usePolling(initialData, url);
+  const getPageUrl = useCallback(
-
+    (cursor: string) => `${firstPageUrl}&cursor=${encodeURIComponent(cursor)}`,
-  const onLimitChange = (next: PageSize) => {
+    [firstPageUrl],
-    router.push(eventsHref({ agent_id: agentId, check_id: checkId, limit: next }));
+  );
-  };
+  const {
+    items,
+    hasMore,
+    loadingMore,
+    error,
+    sentinelRef,
+    hasNewItems,
+    showLatest,
+  } = useInfiniteCursorList<StoredEvent>({
+    initialData,
+    firstPageUrl,
+    getPageUrl,
+    getKey: eventKey,
+    resetKey: `${agentId ?? ""}:${checkId ?? ""}`,
+  });
 
   return (
     <>
-      <PageHeader title="Events" count={data.items.length} />
+      <PageHeader title="Events" count={items.length} />
-      <div className="mb-4 flex items-center justify-between gap-3">
+      <div className="mb-4">
-        <ActiveFilters agentId={agentId} checkId={checkId} limit={limit} />
+        <ActiveFilters agentId={agentId} checkId={checkId} />
-        <PageSizeSelect value={limit} onChange={onLimitChange} />
       </div>
-      {data.items.length === 0 ? (
+      {items.length === 0 ? (
         <EmptyPanel />
       ) : (
         <table className="w-full text-sm">
@@ -68,7 +70,7 @@ export function EventsBrowser({
             </tr>
           </thead>
           <tbody>
-            {data.items.map((e) => (
+            {items.map((e) => (
               <tr key={e.event_id}>
                 <Td>
                   <DateTime iso={e.observed_at} />
@@ -88,12 +90,13 @@ export function EventsBrowser({
           </tbody>
         </table>
       )}
-      <Pager
+      <InfiniteListFooter
-        basePath="/events"
+        sentinelRef={sentinelRef}
-        prevCursor={data.prev_cursor}
+        hasMore={hasMore}
-        nextCursor={data.next_cursor}
+        loading={loadingMore}
-        page={page}
+        error={error}
-        extra={{ agent_id: agentId, check_id: checkId, limit: String(limit) }}
+        hasNewItems={hasNewItems}
+        onShowLatest={showLatest}
       />
     </>
   );
@@ -102,11 +105,9 @@ export function EventsBrowser({
 function ActiveFilters({
   agentId,
   checkId,
-  limit,
 }: {
   agentId?: string;
   checkId?: string;
-  limit: PageSize;
 }) {
   if (!agentId && !checkId) return <div />;
 
@@ -116,18 +117,18 @@ function ActiveFilters({
         <FilterChip
           label="agent_id"
           value={agentId}
-          href={eventsHref({ check_id: checkId, limit })}
+          href={eventsHref({ check_id: checkId })}
         />
       ) : null}
       {checkId ? (
         <FilterChip
           label="check_id"
           value={checkId}
-          href={eventsHref({ agent_id: agentId, limit })}
+          href={eventsHref({ agent_id: agentId })}
         />
       ) : null}
       <Link
-        href={eventsHref({ limit })}
+        href={eventsHref({})}
         className="text-neutral-500 hover:text-neutral-300"
       >
         clear
@@ -153,34 +154,28 @@ function FilterChip({ label, value, href }: { label: string; value: string; href
 function eventsHref(params: {
   agent_id?: string;
   check_id?: string;
-  limit?: PageSize;
 }): string {
   const qs = new URLSearchParams();
   if (params.agent_id) qs.set("agent_id", params.agent_id);
   if (params.check_id) qs.set("check_id", params.check_id);
-  if (params.limit) qs.set("limit", String(params.limit));
   const query = qs.toString();
   return query ? `/events?${query}` : "/events";
 }
 
 function eventsPollUrl({
-  cursor,
-  back,
   agentId,
   checkId,
-  limit,
 }: {
-  cursor?: string;
-  back: boolean;
   agentId?: string;
   checkId?: string;
-  limit: PageSize;
 }) {
   const qs = new URLSearchParams();
-  if (cursor) qs.set("cursor", cursor);
-  if (back) qs.set("back", "1");
   if (agentId) qs.set("agent_id", agentId);
   if (checkId) qs.set("check_id", checkId);
-  qs.set("limit", String(limit));
+  qs.set("limit", String(INFINITE_BATCH_SIZE));
   return `/api/poll/events?${qs.toString()}`;
 }
+
+function eventKey(event: StoredEvent): string {
+  return event.event_id;
+}

web/src/components/IncidentsBrowser.tsx

@@ -2,18 +2,18 @@
 
 import Link from "next/link";
 import { useRouter } from "next/navigation";
-import { useMemo } from "react";
+import { useCallback, useMemo } from "react";
 
 import { AgentLink } from "@/components/AgentLink";
 import { CheckEventsLink } from "@/components/CheckEventsLink";
 import { EmptyPanel, PageHeader, SummaryCell, Td, Th } from "@/components/DataPanel";
-import { PageSizeSelect } from "@/components/PageSizeSelect";
+import { InfiniteListFooter } from "@/components/InfiniteListFooter";
-import { Pager } from "@/components/Pager";
 import { SortHeaderButton, type SortDir } from "@/components/SortHeaderButton";
 import { DateTime } from "@/components/TimeZoneControls";
 import type { components } from "@/lib/api-types";
 import { statusBadgeClass } from "@/lib/format";
-import { type PageSize } from "@/lib/pagination";
+import { INFINITE_BATCH_SIZE } from "@/lib/pagination";
+import { useInfiniteCursorList } from "@/lib/useInfiniteList";
 import { usePolling } from "@/lib/usePolling";
 import type { InventoryData } from "@/lib/view-types";
 
@@ -21,51 +21,56 @@ const STATES = ["all", "open", "resolved"] as const;
 
 type Incident = components["schemas"]["Incident"];
 type Agent = components["schemas"]["Agent"];
-type IncidentsData = { items: Incident[]; next_cursor?: string | null; prev_cursor?: string | null };
+type IncidentsData = { items: Incident[]; next_cursor?: string | null };
 type SortField = "status" | "opened_at" | "resolved_at";
 
 export function IncidentsBrowser({
   initialData,
   initialInventory,
   state,
-  cursor,
-  back,
   sort,
   direction,
-  limit,
-  page,
 }: {
   initialData: IncidentsData;
   initialInventory: InventoryData;
   state?: "open" | "resolved";
-  cursor?: string;
-  back: boolean;
   sort: SortField;
   direction: SortDir;
-  limit: PageSize;
-  page: number;
 }) {
   const router = useRouter();
-  const url = useMemo(() => {
+  const firstPageUrl = useMemo(() => {
     const q = new URLSearchParams();
     if (state) q.set("state", state);
     q.set("sort", sort);
     q.set("direction", direction);
-    q.set("limit", String(limit));
+    q.set("limit", String(INFINITE_BATCH_SIZE));
-    if (cursor) q.set("cursor", cursor);
-    if (back) q.set("back", "1");
     return `/api/poll/incidents?${q.toString()}`;
-  }, [cursor, back, direction, limit, state, sort]);
+  }, [direction, state, sort]);
-  const { data } = usePolling<IncidentsData>(initialData, url);
+  const getPageUrl = useCallback(
+    (cursor: string) => `${firstPageUrl}&cursor=${encodeURIComponent(cursor)}`,
+    [firstPageUrl],
+  );
+  const {
+    items,
+    hasMore,
+    loadingMore,
+    error,
+    sentinelRef,
+    hasNewItems,
+    showLatest,
+  } = useInfiniteCursorList<Incident>({
+    initialData,
+    firstPageUrl,
+    getPageUrl,
+    getKey: incidentKey,
+    resetKey: `${state ?? "all"}:${sort}:${direction}`,
+  });
   const { data: inventory } = usePolling(initialInventory, "/api/poll/inventory");
   const agentsById = useMemo(
     () => new Map(inventory.agents.map((agent) => [agent.agent_id, agent])),
     [inventory.agents],
   );
 
-  const onPageSizeChange = (next: PageSize) => {
-    router.push(incidentsHref({ state, sort, direction, limit: next }));
-  };
   const chooseSort = (field: SortField) => {
     let nextSort = sort;
     let nextDirection = direction;
@@ -75,12 +80,12 @@ export function IncidentsBrowser({
       nextSort = field;
       nextDirection = fieldDefaultDir();
     }
-    router.push(incidentsHref({ state, sort: nextSort, direction: nextDirection, limit }));
+    router.push(incidentsHref({ state, sort: nextSort, direction: nextDirection }));
   };
 
   return (
     <>
-      <PageHeader title="Incidents" count={data.items.length} />
+      <PageHeader title="Incidents" count={items.length} />
       <div className="mb-3 flex items-center justify-between gap-3 text-sm">
         <div className="flex gap-3">
           {STATES.map((s) => {
@@ -88,7 +93,6 @@ export function IncidentsBrowser({
                 state: s === "all" ? undefined : s,
                 sort: sort === "resolved_at" && s !== "resolved" ? "status" : sort,
                 direction,
-              limit,
               });
             const active = (s === "all" && !state) || s === state;
             return (
@@ -102,9 +106,8 @@ export function IncidentsBrowser({
             );
           })}
         </div>
-        <PageSizeSelect value={limit} onChange={onPageSizeChange} />
       </div>
-      {data.items.length === 0 ? (
+      {items.length === 0 ? (
         <EmptyPanel />
       ) : (
         <table className="w-full text-sm">
@@ -150,7 +153,7 @@ export function IncidentsBrowser({
             </tr>
           </thead>
           <tbody>
-            {data.items.map((incident) => (
+            {items.map((incident) => (
               <IncidentRow
                 key={incident.id}
                 incident={incident}
@@ -160,12 +163,13 @@ export function IncidentsBrowser({
           </tbody>
         </table>
       )}
-      <Pager
+      <InfiniteListFooter
-        basePath="/incidents"
+        sentinelRef={sentinelRef}
-        prevCursor={data.prev_cursor}
+        hasMore={hasMore}
-        nextCursor={data.next_cursor}
+        loading={loadingMore}
-        page={page}
+        error={error}
-        extra={{ state, sort, direction, limit: String(limit) }}
+        hasNewItems={hasNewItems}
+        onShowLatest={showLatest}
       />
     </>
   );
@@ -220,6 +224,10 @@ function incidentCheckId(incident: Incident): string {
   return incident.check_id ?? "";
 }
 
+function incidentKey(incident: Incident): string {
+  return String(incident.id);
+}
+
 function fieldDefaultDir(): SortDir {
   return "desc";
 }
@@ -228,18 +236,15 @@ function incidentsHref({
   state,
   sort,
   direction,
-  limit,
 }: {
   state?: "open" | "resolved";
   sort: SortField;
   direction: SortDir;
-  limit: PageSize;
 }) {
   const qs = new URLSearchParams();
   if (state) qs.set("state", state);
   qs.set("sort", sort);
   qs.set("direction", direction);
-  qs.set("limit", String(limit));
   const query = qs.toString();
   return query ? `/incidents?${query}` : "/incidents";
 }

web/src/components/InfiniteListFooter.tsx

@@ -0,0 +1,40 @@
+"use client";
+
+import type { RefObject } from "react";
+
+export function InfiniteListFooter({
+  sentinelRef,
+  hasMore,
+  loading,
+  error,
+  hasNewItems,
+  onShowLatest,
+}: {
+  sentinelRef: RefObject<HTMLDivElement | null>;
+  hasMore: boolean;
+  loading?: boolean;
+  error?: string | null;
+  hasNewItems?: boolean;
+  onShowLatest?: () => void;
+}) {
+  if (!hasMore && !loading && !error && !hasNewItems) return null;
+
+  return (
+    <>
+      {hasNewItems && onShowLatest ? (
+        <button
+          type="button"
+          onClick={onShowLatest}
+          className="fixed bottom-4 right-4 z-40 border border-emerald-800 bg-neutral-950 px-3 py-2 text-xs text-emerald-200 shadow-lg hover:border-emerald-500"
+        >
+          show latest
+        </button>
+      ) : null}
+      <div ref={sentinelRef} className="mt-4 min-h-10 py-2 text-center text-xs text-neutral-500">
+        {error ? <span className="text-red-300">{error}</span> : null}
+        {!error && loading ? "loading..." : null}
+        {!error && !loading && hasMore ? "scroll for more" : null}
+      </div>
+    </>
+  );
+}

web/src/components/ListToolbar.tsx

@@ -1,20 +1,13 @@
 "use client";
 
-import { PageSizeSelect } from "@/components/PageSizeSelect";
-import { type PageSize } from "@/lib/pagination";
-
 export function ListToolbar({
   query,
   onQueryChange,
   placeholder,
-  pageSize,
-  onPageSizeChange,
 }: {
   query: string;
   onQueryChange: (value: string) => void;
   placeholder: string;
-  pageSize: PageSize;
-  onPageSizeChange: (next: PageSize) => void;
 }) {
   return (
     <div className="mb-4 flex items-center justify-between gap-3">
@@ -36,7 +29,6 @@ export function ListToolbar({
           </button>
         ) : null}
       </div>
-      <PageSizeSelect value={pageSize} onChange={onPageSizeChange} />
     </div>
   );
 }

web/src/components/OutboxBrowser.tsx

@@ -1,66 +1,67 @@
 "use client";
 
 import { useRouter } from "next/navigation";
-import { useMemo } from "react";
+import { useCallback, useMemo } from "react";
 
 import { EmptyPanel, PageHeader, Td, Th } from "@/components/DataPanel";
-import { PageSizeSelect } from "@/components/PageSizeSelect";
+import { InfiniteListFooter } from "@/components/InfiniteListFooter";
-import { Pager } from "@/components/Pager";
 import { SortHeaderButton, type SortDir } from "@/components/SortHeaderButton";
 import { DateTime } from "@/components/TimeZoneControls";
 import type { components } from "@/lib/api-types";
 import { statusBadgeClass } from "@/lib/format";
-import { type PageSize } from "@/lib/pagination";
+import { INFINITE_BATCH_SIZE } from "@/lib/pagination";
-import { usePolling } from "@/lib/usePolling";
+import { useInfiniteCursorList } from "@/lib/useInfiniteList";
 
 type OutboxItem = components["schemas"]["NotificationOutboxItem"];
-type OutboxData = { items: OutboxItem[]; next_cursor?: string | null; prev_cursor?: string | null };
+type OutboxData = { items: OutboxItem[]; next_cursor?: string | null };
 type SortField = "created_at" | "updated_at";
 
 export function OutboxBrowser({
   initialData,
-  cursor,
-  back,
   sort,
   direction,
-  limit,
-  page,
 }: {
   initialData: OutboxData;
-  cursor?: string;
-  back: boolean;
   sort: SortField;
   direction: SortDir;
-  limit: PageSize;
-  page: number;
 }) {
   const router = useRouter();
-  const url = useMemo(() => {
+  const firstPageUrl = useMemo(() => {
     const q = new URLSearchParams();
     q.set("sort", sort);
     q.set("direction", direction);
-    q.set("limit", String(limit));
+    q.set("limit", String(INFINITE_BATCH_SIZE));
-    if (cursor) q.set("cursor", cursor);
-    if (back) q.set("back", "1");
     return `/api/poll/outbox?${q.toString()}`;
-  }, [cursor, back, direction, limit, sort]);
+  }, [direction, sort]);
-  const { data } = usePolling<OutboxData>(initialData, url);
+  const getPageUrl = useCallback(
+    (cursor: string) => `${firstPageUrl}&cursor=${encodeURIComponent(cursor)}`,
+    [firstPageUrl],
+  );
+  const {
+    items,
+    hasMore,
+    loadingMore,
+    error,
+    sentinelRef,
+    hasNewItems,
+    showLatest,
+  } = useInfiniteCursorList<OutboxItem>({
+    initialData,
+    firstPageUrl,
+    getPageUrl,
+    getKey: outboxKey,
+    resetKey: `${sort}:${direction}`,
+  });
 
-  const onPageSizeChange = (next: PageSize) => {
-    router.push(outboxHref({ sort, direction, limit: next }));
-  };
   const chooseSort = (field: SortField) => {
     const nextDirection = field === sort && direction === "desc" ? "asc" : "desc";
-    router.push(outboxHref({ sort: field, direction: nextDirection, limit }));
+    router.push(outboxHref({ sort: field, direction: nextDirection }));
   };
 
   return (
     <>
-      <PageHeader title="Notification outbox" count={data.items.length} />
+      <PageHeader title="Notification outbox" count={items.length} />
-      <div className="mb-3 flex justify-end">
+      {items.length === 0 ? (
-        <PageSizeSelect value={limit} onChange={onPageSizeChange} />
-      </div>
-      {data.items.length === 0 ? (
         <EmptyPanel />
       ) : (
         <table className="w-full text-sm">
@@ -80,7 +81,7 @@ export function OutboxBrowser({
             </tr>
           </thead>
           <tbody>
-            {data.items.map((o) => (
+            {items.map((o) => (
               <tr key={o.id}>
                 <Td>
                   <DateTime iso={o.updated_at} />
@@ -98,12 +99,13 @@ export function OutboxBrowser({
           </tbody>
         </table>
       )}
-      <Pager
+      <InfiniteListFooter
-        basePath="/outbox"
+        sentinelRef={sentinelRef}
-        prevCursor={data.prev_cursor}
+        hasMore={hasMore}
-        nextCursor={data.next_cursor}
+        loading={loadingMore}
-        page={page}
+        error={error}
-        extra={{ sort, direction, limit: String(limit) }}
+        hasNewItems={hasNewItems}
+        onShowLatest={showLatest}
       />
     </>
   );
@@ -112,15 +114,16 @@ export function OutboxBrowser({
 function outboxHref({
   sort,
   direction,
-  limit,
 }: {
   sort: SortField;
   direction: SortDir;
-  limit: PageSize;
 }) {
   const qs = new URLSearchParams();
   qs.set("sort", sort);
   qs.set("direction", direction);
-  qs.set("limit", String(limit));
   return `/outbox?${qs.toString()}`;
 }
+
+function outboxKey(item: OutboxItem): string {
+  return String(item.id);
+}

web/src/components/PageSizeSelect.tsx

@@ -1,30 +0,0 @@
-"use client";
-
-import { PAGE_SIZE_OPTIONS, type PageSize } from "@/lib/pagination";
-
-export function PageSizeSelect({
-  value,
-  onChange,
-  options = PAGE_SIZE_OPTIONS,
-}: {
-  value: PageSize;
-  onChange: (size: PageSize) => void;
-  options?: readonly number[];
-}) {
-  return (
-    <label className="inline-flex items-center gap-2 text-xs text-neutral-400">
-      <span>per page</span>
-      <select
-        value={value}
-        onChange={(e) => onChange(Number(e.target.value) as PageSize)}
-        className="border border-neutral-800 bg-neutral-950 px-2 py-1 text-sm text-neutral-100 outline-none focus:border-neutral-500"
-      >
-        {options.map((n) => (
-          <option key={n} value={n}>
-            {n}
-          </option>
-        ))}
-      </select>
-    </label>
-  );
-}

web/src/components/Pager.tsx

@@ -1,61 +0,0 @@
-import Link from "next/link";
-
-export function Pager({
-  basePath,
-  prevCursor,
-  nextCursor,
-  page = 1,
-  extra = {},
-}: {
-  basePath: string;
-  prevCursor?: string | null;
-  nextCursor?: string | null;
-  page?: number;
-  extra?: Record<string, string | undefined>;
-}) {
-  const make = ({
-    nextPage,
-    nextPageCursor,
-    back,
-  }: {
-    nextPage: number;
-    nextPageCursor?: string;
-    back?: boolean;
-  }) => {
-    const sp = new URLSearchParams();
-    for (const [k, v] of Object.entries(extra)) if (v) sp.set(k, v);
-    if (nextPageCursor) sp.set("cursor", nextPageCursor);
-    if (back) sp.set("back", "1");
-    if (nextPage > 1) sp.set("page", String(nextPage));
-    const qs = sp.toString();
-    return qs ? `${basePath}?${qs}` : basePath;
-  };
-
-  const previousHref =
-    page > 1 && prevCursor
-      ? make({ nextPage: page - 1, nextPageCursor: prevCursor, back: true })
-      : null;
-  const nextHref = nextCursor
-    ? make({ nextPage: page + 1, nextPageCursor: nextCursor })
-    : null;
-
-  return (
-    <div className="mt-4 flex justify-between text-sm">
-      {previousHref ? (
-        <Link href={previousHref} className="text-blue-300 hover:text-blue-200">
-          ← prev
-        </Link>
-      ) : (
-        <span className="text-neutral-600">← prev</span>
-      )}
-      <span className="text-neutral-500">page {page}</span>
-      {nextHref ? (
-        <Link href={nextHref} className="text-blue-300 hover:text-blue-200">
-          next →
-        </Link>
-      ) : (
-        <span className="text-neutral-600">end</span>
-      )}
-    </div>
-  );
-}

web/src/components/TimeZoneControls.tsx

@@ -1,10 +1,15 @@
 "use client";
 
-import { createContext, useContext, useMemo, useState } from "react";
+import { createContext, useContext, useEffect, useMemo, useState } from "react";
 
 import { DEFAULT_TIME_ZONE, fmtDate, normalizeTimeZone } from "@/lib/format";
-
+import {
-type TimeZoneMode = "configured" | "browser" | "utc";
+  normalizeTimeZoneMode,
+  TZ_BROWSER_COOKIE,
+  TZ_COOKIE_MAX_AGE_SEC,
+  TZ_MODE_COOKIE,
+  type TimeZoneMode,
+} from "@/lib/timezone-preference";
 
 const modes: TimeZoneMode[] = ["configured", "browser", "utc"];
 
@@ -20,14 +25,29 @@ const TimeZoneContext = createContext<TimeZoneContextValue | null>(null);
 
 export function TimeZoneProvider({
   configuredTimeZone,
+  initialMode,
+  initialBrowserTimeZone,
   children,
 }: {
   configuredTimeZone: string;
+  initialMode: TimeZoneMode;
+  initialBrowserTimeZone: string | null;
   children: React.ReactNode;
 }) {
   const configured = normalizeTimeZone(configuredTimeZone);
-  const [mode, setMode] = useState<TimeZoneMode>("configured");
+  const [mode, setMode] = useState<TimeZoneMode>(normalizeTimeZoneMode(initialMode));
-  const [browserTimeZone, setBrowserTimeZone] = useState<string | null>(null);
+  const [browserTimeZone, setBrowserTimeZone] = useState<string | null>(
+    initialBrowserTimeZone ? normalizeTimeZone(initialBrowserTimeZone) : null,
+  );
+
+  useEffect(() => {
+    if (mode !== "browser") return;
+    const detected = normalizeTimeZone(Intl.DateTimeFormat().resolvedOptions().timeZone);
+    writeCookie(TZ_BROWSER_COOKIE, detected);
+    if (detected === browserTimeZone) return;
+    const id = window.setTimeout(() => setBrowserTimeZone(detected), 0);
+    return () => window.clearTimeout(id);
+  }, [browserTimeZone, mode]);
 
   const value = useMemo<TimeZoneContextValue>(() => {
     const timeZone =
@@ -43,8 +63,13 @@ export function TimeZoneProvider({
       timeZone,
       cycleMode: () => {
         const next = modes[(modes.indexOf(mode) + 1) % modes.length];
+        writeCookie(TZ_MODE_COOKIE, next);
         if (next === "browser" && browserTimeZone === null) {
-          setBrowserTimeZone(normalizeTimeZone(Intl.DateTimeFormat().resolvedOptions().timeZone));
+          const detected = normalizeTimeZone(Intl.DateTimeFormat().resolvedOptions().timeZone);
+          setBrowserTimeZone(detected);
+          writeCookie(TZ_BROWSER_COOKIE, detected);
+        } else if (next !== "browser") {
+          deleteCookie(TZ_BROWSER_COOKIE);
         }
         setMode(next);
       },
@@ -54,6 +79,15 @@ export function TimeZoneProvider({
   return <TimeZoneContext.Provider value={value}>{children}</TimeZoneContext.Provider>;
 }
 
+function writeCookie(name: string, value: string) {
+  const secure = window.location.protocol === "https:" ? "; Secure" : "";
+  document.cookie = `${name}=${encodeURIComponent(value)}; Path=/; SameSite=Lax; Max-Age=${TZ_COOKIE_MAX_AGE_SEC}${secure}`;
+}
+
+function deleteCookie(name: string) {
+  document.cookie = `${name}=; Path=/; SameSite=Lax; Max-Age=0`;
+}
+
 export function DateTime({ iso }: { iso?: string | null }) {
   const { timeZone } = useDisplayTimeZone();
   return <time dateTime={iso ?? undefined}>{fmtDate(iso, timeZone)}</time>;

web/src/components/TopNav.tsx

@@ -18,7 +18,13 @@ const ZERO_SUMMARY: SystemSummaryData = {
   generated_at: "",
 };
 
-export function TopNav({ initialSummary }: { initialSummary: SystemSummaryData | null }) {
+export function TopNav({
+  initialSummary,
+  showLogout,
+}: {
+  initialSummary: SystemSummaryData | null;
+  showLogout: boolean;
+}) {
   const { data } = usePolling(initialSummary ?? ZERO_SUMMARY, "/api/poll/system-summary");
 
   return (
@@ -50,6 +56,16 @@ export function TopNav({ initialSummary }: { initialSummary: SystemSummaryData |
         </Link>
       </nav>
       <TimeZoneCarousel />
+      {showLogout ? (
+        <form action="/api/auth/logout" method="post">
+          <button
+            type="submit"
+            className="rounded border border-neutral-800 bg-neutral-950 px-2.5 py-1.5 text-xs text-neutral-300 hover:border-neutral-600 hover:text-neutral-100"
+          >
+            logout
+          </button>
+        </form>
+      ) : null}
     </header>
   );
 }

web/src/lib/api-types.ts

@@ -327,7 +327,7 @@ export interface components {
         ErrorResponse: {
             error: {
                 /** @enum {string} */
-                code: "unauthorized" | "forbidden" | "conflict" | "agent_blocked" | "not_found" | "validation" | "payload_too_large" | "rate_limited" | "internal" | "not_ready";
+                code: "unauthorized" | "forbidden" | "conflict" | "agent_blocked" | "agent_not_accepted" | "not_found" | "validation" | "payload_too_large" | "rate_limited" | "internal" | "not_ready";
                 message: string;
                 /** @description Server-generated request id, mirrored from X-Request-Id. */
                 request_id: string;
@@ -704,7 +704,7 @@ export interface operations {
             };
             400: components["responses"]["ValidationError"];
             401: components["responses"]["Unauthorized"];
-            /** @description Agent key is blocked. */
+            /** @description Agent key is blocked (`agent_blocked`) or not yet accepted (`agent_not_accepted`). */
             403: {
                 headers: {
                     [name: string]: unknown;

web/src/lib/loaders.ts

@@ -69,48 +69,30 @@ export async function loadEvents({
   agentId,
   checkId,
   cursor,
-  back,
   limit,
 }: {
   agentId?: string;
   checkId?: string;
   cursor?: string;
-  back?: boolean;
   limit?: number;
-} = {}): Promise<{ items: StoredEvent[]; next_cursor?: string | null; prev_cursor?: string | null }> {
+} = {}): Promise<{ items: StoredEvent[]; next_cursor?: string | null }> {
-  return api.queryEvents({ agent_id: agentId, check_id: checkId, cursor, back, limit });
+  return api.queryEvents({ agent_id: agentId, check_id: checkId, cursor, limit });
 }
 
 // PH-001: incidents are unbounded history. Always page through the server.
 export async function loadIncidentsPage(
   q: NonNullable<operations["listIncidents"]["parameters"]["query"]> = {},
-): Promise<{ items: Incident[]; next_cursor?: string | null; prev_cursor?: string | null }> {
+): Promise<{ items: Incident[]; next_cursor?: string | null }> {
   return api.listIncidents(q);
 }
 
 // PH-002: outbox is unbounded. Always page through the server.
 export async function loadOutboxPage(
   q: NonNullable<operations["listOutbox"]["parameters"]["query"]> = {},
-): Promise<{ items: OutboxItem[]; next_cursor?: string | null; prev_cursor?: string | null }> {
+): Promise<{ items: OutboxItem[]; next_cursor?: string | null }> {
   return api.listOutbox(q);
 }
 
-// First-page convenience loaders for SSR. Replace the previous drain-all paths
-// (PH-001, PH-002): they returned the entire table on every initial render and
-// did not scale past a few thousand rows. Pages now show one page and the UI
-// owns subsequent fetches via /api/poll/* with cursors.
-export async function loadIncidents(
-  state?: "open" | "resolved",
-): Promise<Incident[]> {
-  const page = await api.listIncidents({ state, limit: 100 });
-  return page.items;
-}
-
-export async function loadOutboxAll(): Promise<OutboxItem[]> {
-  const page = await api.listOutbox({ limit: 100 });
-  return page.items;
-}
-
 export async function loadSystemSummary(): Promise<SystemSummaryData> {
   return api.getSystemSummary();
 }

web/src/lib/pagination.ts

@@ -1,32 +1,6 @@
-export const PAGE_SIZE_OPTIONS = [10, 50, 100] as const;
+export const INFINITE_BATCH_SIZE = 50;
-export type PageSize = (typeof PAGE_SIZE_OPTIONS)[number];
-export const DEFAULT_PAGE_SIZE: PageSize = 10;
 
-export function pageCountOf(total: number, pageSize: number): number {
+export function normalizeFetchLimit(value: unknown, fallback = INFINITE_BATCH_SIZE): number {
-  return Math.max(1, Math.ceil(total / pageSize));
-}
-
-export function clampPage(page: number, pageCount: number): number {
-  if (page < 1) return 1;
-  if (page > pageCount) return pageCount;
-  return page;
-}
-
-export function paginate<T>(items: T[], page: number, pageSize: number): T[] {
-  const start = (page - 1) * pageSize;
-  return items.slice(start, start + pageSize);
-}
-
-export function normalizePageSize(value: unknown, fallback: PageSize = DEFAULT_PAGE_SIZE): PageSize {
   const n = Number(value);
-  return (PAGE_SIZE_OPTIONS as readonly number[]).includes(n) ? (n as PageSize) : fallback;
+  return Number.isFinite(n) ? Math.min(Math.max(Math.floor(n), 1), 500) : fallback;
-}
-
-export function normalizePageNumber(value: unknown): number {
-  const n = Number(value);
-  return Number.isInteger(n) && n > 0 ? n : 1;
-}
-
-export function normalizeBack(value: unknown): boolean {
-  return value === "1" || value === "true";
 }

web/src/lib/redirect-to-login.ts

@@ -0,0 +1,5 @@
+export function redirectToLogin() {
+  const next = `${window.location.pathname}${window.location.search}`;
+  const params = new URLSearchParams({ next });
+  window.location.assign(`/login?${params.toString()}`);
+}

web/src/lib/timezone-preference.ts

@@ -0,0 +1,9 @@
+export type TimeZoneMode = "configured" | "browser" | "utc";
+
+export const TZ_MODE_COOKIE = "monlet_tz_mode";
+export const TZ_BROWSER_COOKIE = "monlet_tz_browser";
+export const TZ_COOKIE_MAX_AGE_SEC = 31_536_000;
+
+export function normalizeTimeZoneMode(value?: string | null): TimeZoneMode {
+  return value === "browser" || value === "utc" ? value : "configured";
+}

web/src/lib/timezone.ts

@@ -1,7 +1,32 @@
 import "server-only";
 
+import { cookies } from "next/headers";
+
 import { DEFAULT_TIME_ZONE, normalizeTimeZone } from "@/lib/format";
+import {
+  normalizeTimeZoneMode,
+  TZ_BROWSER_COOKIE,
+  TZ_MODE_COOKIE,
+  type TimeZoneMode,
+} from "@/lib/timezone-preference";
 
 export function getUiTimeZone(): string {
   return normalizeTimeZone(process.env.MONLET_WEB_TIME_ZONE ?? process.env.TZ ?? DEFAULT_TIME_ZONE);
 }
+
+export async function getInitialTimeZonePreference(): Promise<{
+  mode: TimeZoneMode;
+  browserTimeZone: string | null;
+}> {
+  const store = await cookies();
+  const mode = normalizeTimeZoneMode(store.get(TZ_MODE_COOKIE)?.value);
+  const browserRaw = store.get(TZ_BROWSER_COOKIE)?.value;
+  const browserTimeZone = browserRaw ? normalizeTimeZone(browserRaw) : null;
+  return {
+    mode,
+    browserTimeZone:
+      browserTimeZone === DEFAULT_TIME_ZONE && browserRaw !== DEFAULT_TIME_ZONE
+        ? null
+        : browserTimeZone,
+  };
+}

web/src/lib/useInfiniteList.ts

@@ -0,0 +1,235 @@
+"use client";
+
+import { useCallback, useEffect, useRef, useState } from "react";
+
+import { redirectToLogin } from "@/lib/redirect-to-login";
+import { usePolling } from "@/lib/usePolling";
+
+export type CursorPage<T> = {
+  items: T[];
+  next_cursor?: string | null;
+};
+
+export function useInfiniteClientList<T>({
+  items,
+  batchSize,
+  resetKey,
+}: {
+  items: T[];
+  batchSize: number;
+  resetKey: string;
+}) {
+  const stateKey = `${resetKey}:${batchSize}`;
+  const [state, setState] = useState({ key: stateKey, visibleCount: batchSize });
+
+  const visibleCount = state.key === stateKey ? state.visibleCount : batchSize;
+
+  const hasMore = visibleCount < items.length;
+  const loadMore = useCallback(() => {
+    setState((current) => {
+      const currentCount = current.key === stateKey ? current.visibleCount : batchSize;
+      return {
+        key: stateKey,
+        visibleCount: Math.min(currentCount + batchSize, items.length),
+      };
+    });
+  }, [batchSize, items.length, stateKey]);
+  const sentinelRef = useInfiniteSentinel(hasMore, loadMore);
+
+  return {
+    visibleItems: items.slice(0, visibleCount),
+    hasMore,
+    sentinelRef,
+    shown: Math.min(visibleCount, items.length),
+    total: items.length,
+  };
+}
+
+export function useInfiniteCursorList<T>({
+  initialData,
+  firstPageUrl,
+  getPageUrl,
+  getKey,
+  resetKey,
+}: {
+  initialData: CursorPage<T>;
+  firstPageUrl: string;
+  getPageUrl: (cursor: string) => string;
+  getKey: (item: T) => string;
+  resetKey: string;
+}) {
+  const { data: firstPage, error, isStale, lastUpdatedAt } = usePolling(initialData, firstPageUrl);
+  const [windowState, setWindowState] = useState(() =>
+    initialWindowState(resetKey, initialData, getKey),
+  );
+  const [loadingMore, setLoadingMore] = useState(false);
+  const [loadError, setLoadError] = useState<string | null>(null);
+  const loadingMoreRef = useRef(false);
+  const abortRef = useRef<AbortController | null>(null);
+  const generationRef = useRef(0);
+  const nearTop = useNearTop();
+  const firstPageSignature = pageSignature(firstPage, getKey);
+  const activeWindow =
+    windowState.key === resetKey ? windowState : initialWindowState(resetKey, initialData, getKey);
+
+  // Persist the head rebase into state (single source of truth) so an in-flight
+  // loadMore cannot glue a stale tail onto an outdated base and revert a fresh
+  // head. Skip while a tail load is in flight: that request extends the current
+  // base and must resolve against it.
+  useEffect(() => {
+    if (loadingMoreRef.current) return;
+    setWindowState((current) => {
+      const w =
+        current.key === resetKey ? current : initialWindowState(resetKey, initialData, getKey);
+      if (w.tailItems.length === 0 && w.baseSignature !== firstPageSignature && nearTop) {
+        return initialWindowState(resetKey, firstPage, getKey);
+      }
+      return w;
+    });
+  }, [firstPageSignature, nearTop, resetKey, firstPage, initialData, getKey]);
+
+  const hasNewHead = activeWindow.baseSignature !== firstPageSignature;
+  const baseItems = activeWindow.baseItems;
+  const tailItems = activeWindow.tailItems;
+  const items = appendUnique(baseItems, tailItems, getKey);
+  const nextCursor = activeWindow.nextCursor;
+
+  const showLatest = useCallback(() => {
+    generationRef.current += 1;
+    abortRef.current?.abort();
+    abortRef.current = null;
+    loadingMoreRef.current = false;
+    setLoadingMore(false);
+    setLoadError(null);
+    setWindowState(initialWindowState(resetKey, firstPage, getKey));
+    if (typeof window !== "undefined") window.scrollTo({ top: 0, behavior: "smooth" });
+  }, [firstPage, getKey, resetKey]);
+
+  const loadMore = useCallback(() => {
+    if (!nextCursor || loadingMoreRef.current) return;
+    loadingMoreRef.current = true;
+    setLoadingMore(true);
+    setLoadError(null);
+    const controller = new AbortController();
+    abortRef.current = controller;
+    const generation = generationRef.current;
+    const requestBaseSignature = activeWindow.baseSignature;
+
+    fetch(getPageUrl(nextCursor), { cache: "no-store", signal: controller.signal })
+      .then(async (response) => {
+        if (response.status === 401) {
+          redirectToLogin();
+          throw new Error("authentication required");
+        }
+        if (!response.ok) {
+          const body = await response.json().catch(() => null);
+          throw new Error(body?.error?.message ?? response.statusText);
+        }
+        return response.json() as Promise<CursorPage<T>>;
+      })
+      .then((page) => {
+        if (generation !== generationRef.current) return;
+        setWindowState((current) => {
+          if (current.key !== resetKey) return current;
+          // The cursor was computed against requestBaseSignature. If a poll
+          // rebased the base in state meanwhile, gluing this tail onto a
+          // different base would skip the rows between the new head and the old
+          // cursor. Drop the response instead; the user reloads via showLatest.
+          if (current.baseSignature !== requestBaseSignature) return current;
+          return {
+            key: resetKey,
+            baseItems: current.baseItems,
+            baseSignature: current.baseSignature,
+            tailItems: appendUnique(current.tailItems, page.items, getKey),
+            nextCursor: page.next_cursor ?? null,
+          };
+        });
+      })
+      .catch((e: unknown) => {
+        if (generation === generationRef.current && !controller.signal.aborted) {
+          setLoadError(e instanceof Error ? e.message : String(e));
+        }
+      })
+      .finally(() => {
+        if (generation !== generationRef.current) return;
+        if (abortRef.current === controller) abortRef.current = null;
+        loadingMoreRef.current = false;
+        setLoadingMore(false);
+      });
+  }, [activeWindow, getKey, getPageUrl, nextCursor, resetKey]);
+
+  useEffect(() => () => abortRef.current?.abort(), []);
+
+  const sentinelRef = useInfiniteSentinel(Boolean(nextCursor), loadMore);
+
+  return {
+    items,
+    hasMore: Boolean(nextCursor),
+    loadingMore,
+    error: loadError ?? error,
+    isStale,
+    lastUpdatedAt,
+    sentinelRef,
+    hasNewItems: hasNewHead,
+    showLatest,
+  };
+}
+
+function useInfiniteSentinel(enabled: boolean, onIntersect: () => void) {
+  const ref = useRef<HTMLDivElement | null>(null);
+  const callbackRef = useRef(onIntersect);
+
+  useEffect(() => {
+    callbackRef.current = onIntersect;
+  }, [onIntersect]);
+
+  useEffect(() => {
+    const node = ref.current;
+    if (!enabled || !node) return;
+    const observer = new IntersectionObserver(
+      (entries) => {
+        if (entries.some((entry) => entry.isIntersecting)) callbackRef.current();
+      },
+      { rootMargin: "360px 0px" },
+    );
+    observer.observe(node);
+    return () => observer.disconnect();
+  }, [enabled]);
+
+  return ref;
+}
+
+function appendUnique<T>(current: T[], next: T[], getKey: (item: T) => string): T[] {
+  const keys = new Set(current.map(getKey));
+  return [...current, ...next.filter((item) => !keys.has(getKey(item)))];
+}
+
+function initialWindowState<T>(key: string, page: CursorPage<T>, getKey: (item: T) => string) {
+  return {
+    key,
+    baseItems: page.items,
+    baseSignature: pageSignature(page, getKey),
+    tailItems: [] as T[],
+    nextCursor: page.next_cursor ?? null,
+  };
+}
+
+function pageSignature<T>(page: CursorPage<T>, getKey: (item: T) => string): string {
+  return `${page.next_cursor ?? ""}:${page.items.map(getKey).join("\u0000")}`;
+}
+
+function useNearTop() {
+  const [nearTop, setNearTop] = useState(true);
+
+  useEffect(() => {
+    const update = () => setNearTop(window.scrollY < 120);
+    const frame = window.requestAnimationFrame(update);
+    window.addEventListener("scroll", update, { passive: true });
+    return () => {
+      window.cancelAnimationFrame(frame);
+      window.removeEventListener("scroll", update);
+    };
+  }, []);
+
+  return nearTop;
+}

web/src/lib/usePolling.ts

@@ -2,6 +2,8 @@
 
 import { useEffect, useState } from "react";
 
+import { redirectToLogin } from "@/lib/redirect-to-login";
+
 const DEFAULT_POLL_MS = 10_000;
 const BACKOFF_AFTER_ERRORS = 2;
 const BACKOFF_CAP_MS = 60_000;
@@ -53,6 +55,10 @@ export function usePolling<T>(initialData: T, url: string, intervalMs = pollInte
 
       fetch(url, { cache: "no-store", signal: controller.signal })
         .then(async (response) => {
+          if (response.status === 401) {
+            redirectToLogin();
+            throw new Error("authentication required");
+          }
           if (!response.ok) {
             const body = await response.json().catch(() => null);
             throw new Error(body?.error?.message ?? response.statusText);

web/src/lib/view-types.ts

@@ -8,13 +8,4 @@ export type InventoryData = {
   checks: CheckState[];
 };
 
-export type SystemSummaryData = {
+export type SystemSummaryData = components["schemas"]["SystemSummaryResponse"];
-  agents_online: number;
-  agents_offline: number;
-  checks_ok: number;
-  checks_warning: number;
-  checks_critical: number;
-  checks_unknown: number;
-  open_incidents: number;
-  generated_at: string;
-};

web/src/lib/web-auth-server.ts

@@ -0,0 +1,52 @@
+import "server-only";
+
+import { cookies, headers } from "next/headers";
+
+import {
+  localAuthEnabled,
+  localAuthMisconfigured,
+  proxyAuthEnabled,
+  SESSION_COOKIE,
+  verifySessionToken,
+} from "@/lib/web-auth";
+
+export async function requestHasAppAccess(): Promise<boolean> {
+  if (localAuthMisconfigured()) return false;
+  if (proxyAuthEnabled()) return Boolean((await headers()).get("x-forwarded-user"));
+  if (!localAuthEnabled()) return true;
+  return verifySessionToken(
+    (await cookies()).get(SESSION_COOKIE)?.value,
+    process.env.MONLET_WEB_SESSION_SECRET ?? "",
+    process.env.MONLET_WEB_AUTH_USERNAME,
+  );
+}
+
+export function showLocalLogout(): boolean {
+  return localAuthEnabled() && !proxyAuthEnabled() && !localAuthMisconfigured();
+}
+
+// CSRF defense for state-changing routes. Prefer the Fetch Metadata signal;
+// fall back to Origin/Host comparison for clients that omit Sec-Fetch-Site.
+export function assertSameOrigin(req: Request): boolean {
+  const fetchSite = req.headers.get("sec-fetch-site");
+  // Strict same-origin: same-site would let a sibling subdomain pass the guard.
+  if (fetchSite) return fetchSite === "same-origin";
+
+  const origin = req.headers.get("origin");
+  if (!origin) return false;
+  let originHost: string;
+  try {
+    originHost = new URL(origin).host;
+  } catch {
+    return false;
+  }
+  return originHost === requestHost(req);
+}
+
+// Trust X-Forwarded-Host like cookieSecure() trusts X-Forwarded-Proto: the app
+// is deployed behind a reverse proxy that terminates the public origin.
+function requestHost(req: Request): string {
+  const forwarded = req.headers.get("x-forwarded-host");
+  if (forwarded) return forwarded.split(",")[0]!.trim();
+  return req.headers.get("host") ?? new URL(req.url).host;
+}

web/src/lib/web-auth.ts

@@ -0,0 +1,170 @@
+export const SESSION_COOKIE = "monlet_session";
+export const DEFAULT_SESSION_TTL_SEC = 604_800;
+export const MIN_SESSION_SECRET_BYTES = 32;
+const LOGIN_FAILURE_DELAY_MS = 350;
+
+type SessionPayload = {
+  u: string;
+  exp: number;
+};
+
+export function localAuthEnabled(): boolean {
+  return Boolean(
+    process.env.MONLET_WEB_AUTH_USERNAME &&
+      process.env.MONLET_WEB_AUTH_PASSWORD &&
+      sessionSecretStrong(process.env.MONLET_WEB_SESSION_SECRET),
+  );
+}
+
+export function localAuthMisconfigured(): boolean {
+  const hasAny = Boolean(
+    process.env.MONLET_WEB_AUTH_USERNAME ||
+      process.env.MONLET_WEB_AUTH_PASSWORD ||
+      process.env.MONLET_WEB_SESSION_SECRET,
+  );
+  const enabled = localAuthEnabled();
+  return (hasAny && !enabled) || (proxyAuthEnabled() && enabled);
+}
+
+export function proxyAuthEnabled(): boolean {
+  return process.env.MONLET_WEB_TRUST_PROXY_AUTH === "true";
+}
+
+export function sessionSecretStrong(secret?: string): boolean {
+  return new TextEncoder().encode(secret ?? "").byteLength >= MIN_SESSION_SECRET_BYTES;
+}
+
+export function sessionTtlSec(): number {
+  const raw = Number(process.env.MONLET_WEB_SESSION_TTL_SEC);
+  return Number.isFinite(raw) && raw > 0 ? Math.floor(raw) : DEFAULT_SESSION_TTL_SEC;
+}
+
+export function sanitizeNext(value?: string | null): string {
+  const fallback = "/agents";
+  const raw = value?.trim();
+  if (!raw || !raw.startsWith("/") || raw.startsWith("//") || raw.includes("\\")) return fallback;
+
+  let decoded: string;
+  try {
+    decoded = decodeURIComponent(raw);
+  } catch {
+    return fallback;
+  }
+  if (
+    !decoded.startsWith("/") ||
+    decoded.startsWith("//") ||
+    decoded.includes("\\") ||
+    hasControlChar(decoded)
+  ) {
+    return fallback;
+  }
+
+  let url: URL;
+  try {
+    url = new URL(raw, "https://monlet.local");
+  } catch {
+    return fallback;
+  }
+  if (url.origin !== "https://monlet.local") return fallback;
+  if (url.pathname === "/login" || url.pathname.startsWith("/api/auth/")) return fallback;
+  return `${url.pathname}${url.search}${url.hash}`;
+}
+
+export function cookieSecure(request: Request): boolean {
+  const forwarded = request.headers.get("x-forwarded-proto");
+  return forwarded === "https" || new URL(request.url).protocol === "https:";
+}
+
+export async function createSessionToken(username: string, secret: string, ttlSec: number) {
+  const payload: SessionPayload = {
+    u: username,
+    exp: Math.floor(Date.now() / 1000) + ttlSec,
+  };
+  const payloadBytes = new TextEncoder().encode(JSON.stringify(payload));
+  const signature = await sign(payloadBytes, secret);
+  return `${base64UrlEncode(payloadBytes)}.${base64UrlEncode(signature)}`;
+}
+
+export async function constantTimeEqualString(actual: string, expected: string): Promise<boolean> {
+  const [actualHash, expectedHash] = await Promise.all([sha256(actual), sha256(expected)]);
+  return equalBytes(actualHash, expectedHash);
+}
+
+export async function delayLoginFailure(): Promise<void> {
+  await new Promise((resolve) => setTimeout(resolve, LOGIN_FAILURE_DELAY_MS));
+}
+
+export async function verifySessionToken(
+  token: string | undefined,
+  secret: string,
+  expectedUsername?: string,
+): Promise<boolean> {
+  if (!token) return false;
+  const [payloadPart, signaturePart] = token.split(".");
+  if (!payloadPart || !signaturePart) return false;
+  const payloadBytes = base64UrlDecode(payloadPart);
+  const signature = base64UrlDecode(signaturePart);
+  if (!payloadBytes || !signature) return false;
+  const expected = await sign(payloadBytes, secret);
+  if (!equalBytes(signature, expected)) return false;
+  try {
+    const payload = JSON.parse(new TextDecoder().decode(payloadBytes)) as SessionPayload;
+    return (
+      typeof payload.u === "string" &&
+      (!expectedUsername || payload.u === expectedUsername) &&
+      payload.exp > Math.floor(Date.now() / 1000)
+    );
+  } catch {
+    return false;
+  }
+}
+
+async function sign(payload: Uint8Array, secret: string): Promise<Uint8Array> {
+  const key = await crypto.subtle.importKey(
+    "raw",
+    new TextEncoder().encode(secret),
+    { name: "HMAC", hash: "SHA-256" },
+    false,
+    ["sign"],
+  );
+  return new Uint8Array(await crypto.subtle.sign("HMAC", key, arrayBufferOf(payload)));
+}
+
+async function sha256(value: string): Promise<Uint8Array> {
+  const bytes = new TextEncoder().encode(value);
+  return new Uint8Array(await crypto.subtle.digest("SHA-256", arrayBufferOf(bytes)));
+}
+
+function arrayBufferOf(bytes: Uint8Array): ArrayBuffer {
+  return bytes.buffer.slice(bytes.byteOffset, bytes.byteOffset + bytes.byteLength) as ArrayBuffer;
+}
+
+function equalBytes(a: Uint8Array, b: Uint8Array): boolean {
+  if (a.length !== b.length) return false;
+  let diff = 0;
+  for (let i = 0; i < a.length; i++) diff |= a[i] ^ b[i];
+  return diff === 0;
+}
+
+function hasControlChar(value: string): boolean {
+  return /[\u0000-\u001F\u007F]/.test(value);
+}
+
+function base64UrlEncode(bytes: Uint8Array): string {
+  return btoa(String.fromCharCode(...bytes))
+    .replaceAll("+", "-")
+    .replaceAll("/", "_")
+    .replaceAll("=", "");
+}
+
+function base64UrlDecode(value: string): Uint8Array | null {
+  try {
+    const padded = value.replaceAll("-", "+").replaceAll("_", "/").padEnd(
+      Math.ceil(value.length / 4) * 4,
+      "=",
+    );
+    return Uint8Array.from(atob(padded), (c) => c.charCodeAt(0));
+  } catch {
+    return null;
+  }
+}

web/src/proxy.ts

@@ -1,29 +1,44 @@
-import type { NextRequest } from "next/server";
+import { NextRequest, NextResponse } from "next/server";
+
+import {
+  localAuthEnabled,
+  localAuthMisconfigured,
+  proxyAuthEnabled,
+  sanitizeNext,
+  SESSION_COOKIE,
+  verifySessionToken,
+} from "@/lib/web-auth";
 
 const ACTION_PATH = "/api/agents/action";
 const HEALTH_PATH = "/api/health";
+const LOGIN_PATH = "/login";
+const AUTH_PREFIX = "/api/auth/";
 
 export const config = {
   matcher: "/((?!_next/static|_next/image|favicon.ico).*)",
 };
 
-export function proxy(request: NextRequest) {
+export async function proxy(request: NextRequest) {
   const path = request.nextUrl.pathname;
-  if (path === HEALTH_PATH) return;
+  if (path === HEALTH_PATH || path === LOGIN_PATH || path.startsWith(AUTH_PREFIX)) return;
 
-  const user = process.env.MONLET_WEB_AUTH_USERNAME;
+  if (localAuthMisconfigured()) {
-  const password = process.env.MONLET_WEB_AUTH_PASSWORD;
+    return new Response("web auth is misconfigured", { status: 500 });
-  if (user && password) {
-    if (validBasicAuth(request, user, password)) return;
-    return new Response("authentication required", {
-      status: 401,
-      headers: { "WWW-Authenticate": 'Basic realm="monlet"' },
-    });
   }
 
-  if (process.env.MONLET_WEB_TRUST_PROXY_AUTH === "true") {
+  if (proxyAuthEnabled()) {
     if (request.headers.get("x-forwarded-user")) return;
-    return new Response("proxy authentication required", { status: 401 });
+    return unauthorized(request, "proxy authentication required");
+  }
+
+  if (localAuthEnabled()) {
+    const ok = await verifySessionToken(
+      request.cookies.get(SESSION_COOKIE)?.value,
+      process.env.MONLET_WEB_SESSION_SECRET ?? "",
+      process.env.MONLET_WEB_AUTH_USERNAME,
+    );
+    if (ok) return;
+    return unauthorized(request, "authentication required");
   }
 
   if (path === ACTION_PATH) {
@@ -34,15 +49,11 @@ export function proxy(request: NextRequest) {
   }
 }
 
-function validBasicAuth(request: NextRequest, user: string, password: string): boolean {
+function unauthorized(request: NextRequest, message: string) {
-  const header = request.headers.get("authorization");
+  if (request.nextUrl.pathname.startsWith("/api/")) {
-  if (!header?.startsWith("Basic ")) return false;
+    return Response.json({ error: { code: "unauthorized", message } }, { status: 401 });
-  try {
-    const raw = atob(header.slice("Basic ".length));
-    const split = raw.indexOf(":");
-    if (split < 0) return false;
-    return raw.slice(0, split) === user && raw.slice(split + 1) === password;
-  } catch {
-    return false;
   }
+  const login = new URL(LOGIN_PATH, request.url);
+  login.searchParams.set("next", sanitizeNext(`${request.nextUrl.pathname}${request.nextUrl.search}`));
+  return NextResponse.redirect(login, { status: 303 });
 }

web/tests/mock-server.mjs

@@ -3,6 +3,60 @@ import { createServer } from "node:http";
 const PORT = Number(process.env.MOCK_PORT ?? 8765);
 
 const now = new Date().toISOString();
+const eventItems = [
+  {
+    event_id: "00000000-0000-7000-8000-000000000001",
+    agent_id: "agent-1",
+    check_id: "disk",
+    observed_at: now,
+    received_at: now,
+    status: "ok",
+    exit_code: 0,
+    duration_ms: 12,
+    output: "disk ok",
+    output_truncated: false,
+    notifications_enabled: true,
+  },
+  {
+    event_id: "00000000-0000-7000-8000-000000000002",
+    agent_id: "agent-1",
+    check_id: "load",
+    observed_at: now,
+    received_at: now,
+    status: "warning",
+    exit_code: 1,
+    duration_ms: 34,
+    output: "load high",
+    output_truncated: false,
+    notifications_enabled: true,
+  },
+  {
+    event_id: "00000000-0000-7000-8000-000000000003",
+    agent_id: "agent-2",
+    check_id: "agent_liveness",
+    observed_at: now,
+    received_at: now,
+    status: "critical",
+    exit_code: 2,
+    duration_ms: 0,
+    output: "agent is dead",
+    output_truncated: false,
+    notifications_enabled: true,
+  },
+  ...Array.from({ length: 12 }, (_, i) => ({
+    event_id: `00000000-0000-7000-8000-0000000001${String(i).padStart(2, "0")}`,
+    agent_id: "agent-1",
+    check_id: "disk",
+    observed_at: now,
+    received_at: now,
+    status: "ok",
+    exit_code: 0,
+    duration_ms: 10 + i,
+    output: `disk extra ${i + 1}`,
+    output_truncated: false,
+    notifications_enabled: true,
+  })),
+];
 
 const fixtures = {
   "/api/v1/health": { status: "ok" },
@@ -12,6 +66,7 @@ const fixtures = {
     checks_ok: 2,
     checks_warning: 1,
     checks_critical: 1,
+    checks_unknown: 0,
     open_incidents: 2,
     generated_at: now,
   },
@@ -128,48 +183,8 @@ const fixtures = {
     next_cursor: null,
   },
   "/api/v1/events/query": {
-    items: [
+    items: eventItems,
-      {
+    next_cursor: "10",
-        event_id: "00000000-0000-7000-8000-000000000001",
-        agent_id: "agent-1",
-        check_id: "disk",
-        observed_at: now,
-        received_at: now,
-        status: "ok",
-        exit_code: 0,
-        duration_ms: 12,
-        output: "disk ok",
-        output_truncated: false,
-        notifications_enabled: true,
-      },
-      {
-        event_id: "00000000-0000-7000-8000-000000000002",
-        agent_id: "agent-1",
-        check_id: "load",
-        observed_at: now,
-        received_at: now,
-        status: "warning",
-        exit_code: 1,
-        duration_ms: 34,
-        output: "load high",
-        output_truncated: false,
-        notifications_enabled: true,
-      },
-      {
-        event_id: "00000000-0000-7000-8000-000000000003",
-        agent_id: "agent-2",
-        check_id: "agent_liveness",
-        observed_at: now,
-        received_at: now,
-        status: "critical",
-        exit_code: 2,
-        duration_ms: 0,
-        output: "agent is dead",
-        output_truncated: false,
-        notifications_enabled: true,
-      },
-    ],
-    next_cursor: null,
   },
   "/api/v1/notifiers/outbox": {
     items: [
@@ -205,15 +220,12 @@ const server = createServer((req, res) => {
   if (url.pathname === "/api/v1/events/query" && body) {
     const agentId = url.searchParams.get("agent_id");
     const checkId = url.searchParams.get("check_id");
-    body = {
+    const items = body.items.filter((item) => {
-      ...body,
-      items: body.items.filter((item) => {
       if (agentId && item.agent_id !== agentId) return false;
       if (checkId && item.check_id !== checkId) return false;
       return true;
-      }),
+    });
-      next_cursor: null,
+    body = paginate(body, items, url);
-    };
   }
   if (url.pathname === "/api/v1/incidents" && body) {
     const state = url.searchParams.get("state");
@@ -239,3 +251,16 @@ const server = createServer((req, res) => {
 server.listen(PORT, "127.0.0.1", () => {
   console.log(`mock server on ${PORT}`);
 });
+
+function paginate(body, items, url) {
+  const limit = Number(url.searchParams.get("limit") ?? 100);
+  const start = Number(url.searchParams.get("cursor") ?? 0);
+  const safeLimit = Number.isFinite(limit) && limit > 0 ? limit : 100;
+  const safeStart = Number.isFinite(start) && start > 0 ? start : 0;
+  const next = safeStart + safeLimit;
+  return {
+    ...body,
+    items: items.slice(safeStart, next),
+    next_cursor: next < items.length ? String(next) : null,
+  };
+}

web/tests/smoke.spec.ts

@@ -107,6 +107,14 @@ test("events page", async ({ page }) => {
   await expect(page.getByRole("link", { name: /check_id=.*disk/ })).toBeVisible();
 });
 
+test("events page loads the next cursor page on scroll", async ({ page }) => {
+  await page.goto("/events");
+  await expect(page.locator("tbody")).toContainText("disk extra 7");
+  await page.evaluate(() => window.scrollTo(0, document.body.scrollHeight));
+  await expect(page.locator("tbody")).toContainText("disk extra 11");
+  await expect(page).toHaveURL(/\/events$/);
+});
+
 test("outbox page", async ({ page }) => {
   await page.goto("/outbox");
   await expect(page.locator("body")).toContainText("debug");