.gitignore

@@ -27,7 +27,7 @@ go.work
 # End of https://www.toptal.com/developers/gitignore/api/go
 
 /uptop
-uptop.db
+uptop.db*
 
 .ssh
 

CHANGELOG.md

@@ -1,46 +1,88 @@
 # Changelog
 
-## [2026.05.2] — 2026-05-23
+## [2026.05.5] — 2026-05-29
 
 ### Added
-- Comprehensive test suite (94 tests across monitor, server, cluster)
-- golangci-lint config with CI enforcement
-- Gitea Actions CI pipeline (test + lint)
-- Graceful shutdown for HTTP and SSH servers
-- Context-aware alert delivery with timeout
-- Request size limits on all POST endpoints
-- Constant-time secret comparison
-- Check interval jitter to prevent thundering herd
-- `--version` flag with build metadata injection
+- Error reason display when monitors go DOWN (#33)
+- Push monitor lifecycle — PENDING, LATE, DOWN states (#34)
+- Logs tab overhaul — severity tags, filtering, recovery durations (#35)
+- Alert channel health indicator and test alerts (#36)
 
-### Fixed
-- Silent JSON unmarshal failures in alert settings
-- Panic on crypto/rand failure replaced with error return
-- Alert delivery errors now logged instead of swallowed
-- log.Fatalf in goroutines replaced with log.Printf
-- Deprecated LineUp/LineDown API calls
+### Changed
+- Visual polish — detail sections, column headers, alert detail (#37)
+
+## [2026.05.4] — 2026-05-27
+
+### Added
+- SSH user seeding from `UPTOP_ADMIN_KEY` env var and `UPTOP_KEYS` file (#31)
+- GoReleaser for binary releases
+- govulncheck in CI pipeline
+- Multi-arch Docker builds (amd64 + arm64)
+
+### Changed
+- CI overhaul — Go 1.26, build caching, streamlined pipeline (#30)
+- Bumped golang.org/x/crypto v0.47.0 → v0.52.0
+- Bumped Alpine 3.21 → 3.23
 
 ### Security
-- Cluster secret compared with crypto/subtle (timing-safe)
-- http.MaxBytesReader on all JSON endpoints
-- ReadHeaderTimeout added to HTTP server
+- Phase 1: SSRF protection, input validation, safe dial (#26)
+- Phase 2: TLS hardening, auth bypass fixes, rate limiting (#27)
+- Phase 3: Graceful degradation, connection limits, timeout enforcement (#28)
+- Phase 4: Code quality, error handling, linter fixes (#29)
 
-## [2026.05.1] — 2026-05-14
+## [2026.05.3] — 2026-05-25
+
+### Added
+- Theme system with 5 dark palettes — Default, Dracula, Nord, Tokyo Night, Gruvbox (#24)
+- `--version` flag with build metadata injection
+- Gitea Actions CI pipeline — test + lint (#20)
+- golangci-lint configuration
+- Comprehensive test suite — 94 tests across monitor, server, cluster (#19)
+- CONTRIBUTING.md and SECURITY.md
+
+### Changed
+- Renamed project from go-upkeep to uptop (#25)
+- Updated LICENSE with dual copyright for independent fork
+
+### Fixed
+- Form validators scoped to relevant monitor types (#23)
+- Graceful shutdown for HTTP, SSH servers and database (#19)
+- Constant-time secret comparison, request size limits (#19)
+- Check interval jitter to prevent thundering herd (#19)
+- TUI visual polish — zebra striping, group icons, sparkline stats (#18)
+
+## [2026.05.2] — 2026-05-22
+
+### Added
+- Incident management and maintenance windows (#17)
+- Production docker-compose.yml
+
+### Fixed
+- Viewport sizing and dynamic chrome calculation (#16)
+- Form height constrained to terminal with resize forwarding
+- Maintenance'd monitors excluded from down count and pulse
+- Group status correctly skips children in maintenance
+
+## [2026.05.1] — 2026-05-16
 
 ### Added
 - Distributed probing with leader + probe nodes
-- Config-as-code (YAML apply/export with dry-run, prune)
-- TUI visual polish (zebra striping, sparklines, breadcrumbs)
-- Incident management and maintenance windows
-- 9 alert providers (Discord, Slack, Email, Ntfy, Telegram, PagerDuty, Pushover, Gotify, Webhook)
+- Config-as-code — YAML apply/export with dry-run and prune
+- TUI polish — status bar, tab badges, detail panel, modals
+- DOWN-first sort, health pulse, site filter
+- Type icons in sites table
+- Sparkline history graphs
+- Persistent state — uptime, status, latency, and logs survive restarts
+- Push token stripping from /status/json response
 
-## [2026.04.1] — Initial independent fork
+## [2026.04.1] — 2026-04-01
 
 ### Added
-- SSH-accessible TUI (Bubble Tea + Wish)
-- 6 check types (HTTP, Push, Ping, Port, DNS, Group)
+- SSH-accessible TUI built on Bubble Tea + Wish
+- 6 check types — HTTP, Push, Ping, Port, DNS, Group
+- 9 alert providers — Discord, Slack, Email, Ntfy, Telegram, PagerDuty, Pushover, Gotify, Webhook
 - SQLite and PostgreSQL support
 - HA clustering with automatic failover
-- Prometheus metrics endpoint
-- Public status page
-- Uptime Kuma import
+- Prometheus /metrics endpoint
+- Public status page (HTML + JSON)
+- Uptime Kuma backup import

README.md

@@ -1,19 +1,49 @@
-# uptop
+<div align="center">
+  <h1>uptop</h1>
+  <p>Self-hosted uptime monitoring with a TUI over SSH.</p>
+  <p>No browser. No client install. Just <code>ssh -p 23234 your-server</code>.</p>
 
-Self-hosted uptime monitor with a TUI you can access over SSH. No browser, no install on the client — just `ssh -p 23234 your-server`.
+  <p>
+    <img src="https://img.shields.io/badge/license-MIT-blue" alt="MIT License">
+    <img src="https://img.shields.io/badge/go-1.26-00ADD8?logo=go&logoColor=white" alt="Go 1.26">
+    <img src="https://img.shields.io/docker/pulls/lerkolabs/uptop" alt="Docker Pulls">
+  </p>
 
-Built on the foundation of [RDGames/go-upkeep](https://github.com/RDGames/go-upkeep).
+  <img src="assets/monitors.png" alt="uptop monitors view" width="800">
+</div>
 
-## What it does
+## What is this
 
-- **6 check types**: HTTP, Push (heartbeat), Ping, Port, DNS, Groups
-- **9 alert providers**: Discord, Slack, Email, Ntfy, Webhook, Telegram, PagerDuty, Pushover, Gotify
-- **Config as code**: define monitors in YAML, apply declaratively, version control your setup
-- **HA clustering**: leader/follower with automatic failover
-- **Prometheus metrics**: `/metrics` endpoint for Grafana dashboards
-- **Public status page**: HTML + JSON, toggle with an env var
-- **SQLite or Postgres**: SQLite for single-node, Postgres for production
-- **Uptime Kuma import**: migrate from Kuma with one command
+An uptime monitor you manage entirely from the terminal. It runs as a server, exposes an SSH endpoint, and drops you into a full TUI — monitors, alerts, logs, nodes, all there.
+
+Built on [RDGames/go-upkeep](https://github.com/RDGames/go-upkeep). Rewritten for clustering, config-as-code, and a proper dashboard.
+
+## Features
+
+- **6 check types** — HTTP, Push (heartbeat), Ping, Port, DNS, Groups
+- **9 alert providers** — Discord, Slack, Email, Ntfy, Webhook, Telegram, PagerDuty, Pushover, Gotify
+- **Config as code** — define monitors in YAML, apply declaratively, version control your setup
+- **HA clustering** — leader/follower with automatic failover
+- **Prometheus metrics** — `/metrics` endpoint, wire it straight to Grafana
+- **Public status page** — HTML + JSON, toggle with an env var
+- **SQLite or Postgres** — SQLite for single-node, Postgres for production
+- **Uptime Kuma import** — migrate from Kuma with one command
+
+## Screenshots
+
+<table>
+  <tr>
+    <td><img src="assets/detail.png" alt="detail panel" width="400"></td>
+    <td><img src="assets/alerts.png" alt="alerts view" width="400"></td>
+  </tr>
+  <tr>
+    <td><img src="assets/logs.png" alt="logs view" width="400"></td>
+    <td><img src="assets/nodes.png" alt="cluster nodes" width="400"></td>
+  </tr>
+  <tr>
+    <td colspan="2" align="center"><img src="assets/theme.png" alt="theme selection" width="600"></td>
+  </tr>
+</table>
 
 ## Quick start
 
@@ -22,7 +52,7 @@ go run cmd/uptop/main.go
 ssh -p 23234 localhost
 ```
 
-Seed some demo data to see it in action:
+Want some data to look at first:
 
 ```bash
 go run cmd/uptop/main.go -demo
@@ -30,22 +60,45 @@ go run cmd/uptop/main.go -demo
 
 ## Install
 
-### From source
+<details>
+<summary><strong>Docker (recommended)</strong></summary>
+
+```yaml
+services:
+  uptop:
+    image: lerkolabs/uptop:latest
+    restart: unless-stopped
+    ports:
+      - "23234:23234"
+      - "8080:8080"
+    environment:
+      - UPTOP_DB_TYPE=sqlite
+      - UPTOP_DB_DSN=/data/uptop.db
+      - UPTOP_STATUS_ENABLED=true
+      # - UPTOP_ADMIN_KEY=ssh-ed25519 AAAA... you@host
+    volumes:
+      - ./data:/data
+```
+
+First run: set `UPTOP_ADMIN_KEY` to your SSH public key, or attach to the container and add it in the Users tab.
+
+</details>
+
+<details>
+<summary><strong>Binary</strong></summary>
+
+Download from [Releases](https://gitea.lerkolabs.com/lerko/uptop/releases).
+
+</details>
+
+<details>
+<summary><strong>From source</strong></summary>
 
 ```bash
 go install gitea.lerkolabs.com/lerko/uptop/cmd/uptop@latest
 ```
 
-### Docker
-
-```bash
-docker pull lerko/uptop:latest
-docker run -p 23234:23234 -p 8080:8080 -v ./data:/data lerko/uptop
-```
-
-### Binary
-
-Download from [Releases](https://gitea.lerkolabs.com/lerko/uptop/releases).
+</details>
 
 ## Config as code
 
@@ -63,35 +116,11 @@ uptop apply -f monitors.yaml --dry-run   # see what would change
 uptop apply -f monitors.yaml --prune     # delete anything not in the YAML
 ```
 
-See [docs/config-as-code.md](docs/config-as-code.md) for the full reference.
-
-## Docker
-
-```yaml
-services:
-  monitor:
-    build: .
-    restart: unless-stopped
-    stdin_open: true
-    tty: true
-    ports:
-      - "23234:23234"
-      - "8080:8080"
-    volumes:
-      - ./data:/data
-      - ./ssh_keys:/app/.ssh
-    environment:
-      - UPTOP_DB_TYPE=sqlite
-      - UPTOP_DB_DSN=/data/uptop.db
-      - UPTOP_STATUS_ENABLED=true
-      - UPTOP_CLUSTER_SECRET=change-me
-```
-
-First run: attach to the container (`docker attach uptop`), go to the Users tab, add your SSH public key. Then detach with `Ctrl+P, Ctrl+Q` and connect normally over SSH.
+Full reference in [docs/config-as-code.md](docs/config-as-code.md).
 
 ## Environment variables
 
-| Variable | Default | What it does |
+| Variable | Default | Description |
 |---|---|---|
 | `UPTOP_PORT` | `23234` | SSH server port |
 | `UPTOP_HTTP_PORT` | `8080` | HTTP server port (status page, push, metrics) |
@@ -103,6 +132,7 @@ First run: attach to the container (`docker attach uptop`), go to the Users tab,
 | `UPTOP_PEER_URL` | | Leader URL for follower nodes |
 | `UPTOP_CLUSTER_SECRET` | | Shared key for cluster + API auth |
 | `UPTOP_INSECURE_SKIP_VERIFY` | `false` | Skip TLS verification for checks |
+| `UPTOP_ADMIN_KEY` | | SSH public key seeded as first admin on startup |
 
 ## Migrating from Uptime Kuma
 

cmd/uptop/main.go

@@ -385,6 +385,7 @@ func runServe(args []string) {
 
 	eng.InitHistory()
 	eng.InitLogs()
+	eng.InitAlertHealth()
 	eng.Start(ctx)
 
 	tlsCert := os.Getenv("UPTOP_TLS_CERT")

internal/cluster/cluster_test.go

@@ -53,6 +53,10 @@ func (m *mockStore) GetNode(string) (models.ProbeNode, error) { return models.Pr
 func (m *mockStore) GetAllNodes() ([]models.ProbeNode, error) { return nil, nil }
 func (m *mockStore) UpdateNodeLastSeen(string) error          { return nil }
 func (m *mockStore) DeleteNode(string) error                  { return nil }
+func (m *mockStore) LoadAlertHealth() (map[int]models.AlertHealthRecord, error) {
+	return nil, nil
+}
+func (m *mockStore) SaveAlertHealth(models.AlertHealthRecord) error { return nil }
 func (m *mockStore) SaveLog(string) error                           { return nil }
 func (m *mockStore) LoadLogs(int) ([]string, error)                 { return nil, nil }
 func (m *mockStore) GetActiveMaintenanceWindows() ([]models.MaintenanceWindow, error) {

internal/metrics/prometheus_test.go

@@ -51,6 +51,10 @@ func (m *mockStore) GetNode(string) (models.ProbeNode, error)         { return m
 func (m *mockStore) GetAllNodes() ([]models.ProbeNode, error)         { return nil, nil }
 func (m *mockStore) UpdateNodeLastSeen(string) error                  { return nil }
 func (m *mockStore) DeleteNode(string) error                          { return nil }
+func (m *mockStore) LoadAlertHealth() (map[int]models.AlertHealthRecord, error) {
+	return nil, nil
+}
+func (m *mockStore) SaveAlertHealth(models.AlertHealthRecord) error { return nil }
 func (m *mockStore) SaveLog(string) error                           { return nil }
 func (m *mockStore) LoadLogs(int) ([]string, error)                 { return nil, nil }
 func (m *mockStore) GetActiveMaintenanceWindows() ([]models.MaintenanceWindow, error) {

internal/models/models.go

@@ -79,6 +79,17 @@ type ProbeNode struct {
 	Version  string
 }
 
+// AlertHealthRecord is the persisted send health of an alert channel. It lets the
+// "last sent" / health indicators survive restarts instead of resetting to "never".
+type AlertHealthRecord struct {
+	AlertID    int
+	LastSendAt time.Time
+	LastSendOK bool
+	LastError  string
+	SendCount  int
+	FailCount  int
+}
+
 type MaintenanceWindow struct {
 	ID          int
 	MonitorID   int

internal/monitor/monitor.go

@@ -146,6 +146,26 @@ func (e *Engine) InitLogs() {
 	e.logStore = logs
 }
 
+// InitAlertHealth restores persisted alert send health so the dashboard shows real
+// "last sent" / health state on startup instead of resetting every channel to "never".
+func (e *Engine) InitAlertHealth() {
+	records, err := e.db.LoadAlertHealth()
+	if err != nil {
+		return
+	}
+	e.alertHealthMu.Lock()
+	defer e.alertHealthMu.Unlock()
+	for id, r := range records {
+		e.alertHealth[id] = AlertHealth{
+			LastSendAt: r.LastSendAt,
+			LastSendOK: r.LastSendOK,
+			LastError:  r.LastError,
+			SendCount:  r.SendCount,
+			FailCount:  r.FailCount,
+		}
+	}
+}
+
 func (e *Engine) GetLogs() []string {
 	e.logMu.RLock()
 	defer e.logMu.RUnlock()
@@ -612,6 +632,18 @@ func (e *Engine) recordAlertResult(alertID int, ok bool, errMsg string) {
 		h.FailCount++
 	}
 	e.alertHealth[alertID] = h
+
+	// Persist best-effort so health survives restarts; DB IO off the alert path.
+	go func(rec models.AlertHealthRecord) {
+		_ = e.db.SaveAlertHealth(rec)
+	}(models.AlertHealthRecord{
+		AlertID:    alertID,
+		LastSendAt: h.LastSendAt,
+		LastSendOK: h.LastSendOK,
+		LastError:  h.LastError,
+		SendCount:  h.SendCount,
+		FailCount:  h.FailCount,
+	})
 }
 
 func (e *Engine) GetAlertHealth(alertID int) AlertHealth {

internal/monitor/monitor_test.go

@@ -63,6 +63,10 @@ func (m *mockStore) GetNode(string) (models.ProbeNode, error)         { return m
 func (m *mockStore) GetAllNodes() ([]models.ProbeNode, error)         { return nil, nil }
 func (m *mockStore) UpdateNodeLastSeen(string) error                  { return nil }
 func (m *mockStore) DeleteNode(string) error                          { return nil }
+func (m *mockStore) LoadAlertHealth() (map[int]models.AlertHealthRecord, error) {
+	return nil, nil
+}
+func (m *mockStore) SaveAlertHealth(models.AlertHealthRecord) error { return nil }
 func (m *mockStore) GetActiveMaintenanceWindows() ([]models.MaintenanceWindow, error) {
 	return nil, nil
 }

internal/server/server_test.go

@@ -65,6 +65,10 @@ func (m *mockStore) AddAlertReturningID(string, string, map[string]string) (int,
 func (m *mockStore) GetAllNodes() ([]models.ProbeNode, error) { return nil, nil }
 func (m *mockStore) UpdateNodeLastSeen(string) error          { return nil }
 func (m *mockStore) DeleteNode(string) error                  { return nil }
+func (m *mockStore) LoadAlertHealth() (map[int]models.AlertHealthRecord, error) {
+	return nil, nil
+}
+func (m *mockStore) SaveAlertHealth(models.AlertHealthRecord) error { return nil }
 func (m *mockStore) SaveLog(string) error                           { return nil }
 func (m *mockStore) LoadLogs(int) ([]string, error)                 { return nil, nil }
 func (m *mockStore) GetAllMaintenanceWindows(int) ([]models.MaintenanceWindow, error) {

internal/store/dialect.go

@@ -14,6 +14,7 @@ type Dialect interface {
 	ImportWipe(tx *sql.Tx)
 	ImportResetSequences(tx *sql.Tx)
 	UpsertNodeSQL() string
+	UpsertAlertHealthSQL() string
 }
 
 func rewritePlaceholders(query string, dollarStyle bool) string {

internal/store/postgres.go

@@ -81,6 +81,14 @@ func (d *PostgresDialect) CreateTablesSQL() []string {
 			changed_at TIMESTAMP DEFAULT NOW()
 		)`,
 		`CREATE INDEX IF NOT EXISTS idx_state_changes_site ON state_changes(site_id, changed_at DESC)`,
+		`CREATE TABLE IF NOT EXISTS alert_health (
+			alert_id INTEGER PRIMARY KEY,
+			last_send_at TIMESTAMP,
+			last_send_ok BOOLEAN DEFAULT FALSE,
+			last_error TEXT DEFAULT '',
+			send_count INTEGER DEFAULT 0,
+			fail_count INTEGER DEFAULT 0
+		)`,
 	}
 }
 
@@ -106,6 +114,10 @@ func (d *PostgresDialect) UpsertNodeSQL() string {
 	return "INSERT INTO nodes (id, name, region, last_seen, version) VALUES ($1, $2, $3, NOW(), $4) ON CONFLICT (id) DO UPDATE SET name = EXCLUDED.name, region = EXCLUDED.region, last_seen = NOW(), version = EXCLUDED.version"
 }
 
+func (d *PostgresDialect) UpsertAlertHealthSQL() string {
+	return "INSERT INTO alert_health (alert_id, last_send_at, last_send_ok, last_error, send_count, fail_count) VALUES ($1, $2, $3, $4, $5, $6) ON CONFLICT (alert_id) DO UPDATE SET last_send_at = EXCLUDED.last_send_at, last_send_ok = EXCLUDED.last_send_ok, last_error = EXCLUDED.last_error, send_count = EXCLUDED.send_count, fail_count = EXCLUDED.fail_count"
+}
+
 func (d *PostgresDialect) ResetSequenceOnEmpty(db *sql.DB, table string) {}
 
 func (d *PostgresDialect) ImportWipe(tx *sql.Tx) {

internal/store/sqlite.go

@@ -88,6 +88,14 @@ func (d *SQLiteDialect) CreateTablesSQL() []string {
 			changed_at DATETIME DEFAULT CURRENT_TIMESTAMP
 		)`,
 		`CREATE INDEX IF NOT EXISTS idx_state_changes_site ON state_changes(site_id, changed_at DESC)`,
+		`CREATE TABLE IF NOT EXISTS alert_health (
+			alert_id INTEGER PRIMARY KEY,
+			last_send_at DATETIME,
+			last_send_ok BOOLEAN DEFAULT 0,
+			last_error TEXT DEFAULT '',
+			send_count INTEGER DEFAULT 0,
+			fail_count INTEGER DEFAULT 0
+		)`,
 	}
 }
 
@@ -113,6 +121,10 @@ func (d *SQLiteDialect) UpsertNodeSQL() string {
 	return "INSERT OR REPLACE INTO nodes (id, name, region, last_seen, version) VALUES (?, ?, ?, CURRENT_TIMESTAMP, ?)"
 }
 
+func (d *SQLiteDialect) UpsertAlertHealthSQL() string {
+	return "INSERT OR REPLACE INTO alert_health (alert_id, last_send_at, last_send_ok, last_error, send_count, fail_count) VALUES (?, ?, ?, ?, ?, ?)"
+}
+
 func (d *SQLiteDialect) ResetSequenceOnEmpty(db *sql.DB, table string) {
 	var count int
 	_ = db.QueryRow("SELECT COUNT(*) FROM " + table).Scan(&count) //nolint:errcheck

internal/store/sqlstore.go

@@ -430,6 +430,37 @@ func (s *SQLStore) DeleteNode(id string) error {
 	return err
 }
 
+func (s *SQLStore) LoadAlertHealth() (map[int]models.AlertHealthRecord, error) {
+	rows, err := s.db.Query("SELECT alert_id, last_send_at, last_send_ok, last_error, send_count, fail_count FROM alert_health")
+	if err != nil {
+		return nil, err
+	}
+	defer rows.Close()
+	out := make(map[int]models.AlertHealthRecord)
+	for rows.Next() {
+		var r models.AlertHealthRecord
+		var lastSend sql.NullTime
+		if err := rows.Scan(&r.AlertID, &lastSend, &r.LastSendOK, &r.LastError, &r.SendCount, &r.FailCount); err != nil {
+			return out, err
+		}
+		if lastSend.Valid {
+			r.LastSendAt = lastSend.Time
+		}
+		out[r.AlertID] = r
+	}
+	return out, rows.Err()
+}
+
+func (s *SQLStore) SaveAlertHealth(h models.AlertHealthRecord) error {
+	var lastSend interface{}
+	if !h.LastSendAt.IsZero() {
+		lastSend = h.LastSendAt
+	}
+	_, err := s.db.Exec(s.dialect.UpsertAlertHealthSQL(),
+		h.AlertID, lastSend, h.LastSendOK, h.LastError, h.SendCount, h.FailCount)
+	return err
+}
+
 func (s *SQLStore) SaveLog(message string) error {
 	_, err := s.db.Exec(s.q("INSERT INTO logs (message) VALUES (?)"), message)
 	if err != nil {

internal/store/store.go

@@ -49,6 +49,10 @@ type Store interface {
 	UpdateNodeLastSeen(id string) error
 	DeleteNode(id string) error
 
+	// Alert Health
+	LoadAlertHealth() (map[int]models.AlertHealthRecord, error)
+	SaveAlertHealth(h models.AlertHealthRecord) error
+
 	// Logs
 	SaveLog(message string) error
 	LoadLogs(limit int) ([]string, error)

internal/tui/tab_sites.go

@@ -60,14 +60,18 @@ type siteFormData struct {
 	Regions       string
 }
 
-func latencySparkline(latencies []time.Duration, width int) string {
+func latencySparkline(latencies []time.Duration, statuses []bool, width int) string {
 	if len(latencies) == 0 {
 		return subtleStyle.Render(strings.Repeat("·", width))
 	}
 
 	samples := latencies
+	sampledStatuses := statuses
 	if len(samples) > width {
 		samples = samples[len(samples)-width:]
+		if len(sampledStatuses) > width {
+			sampledStatuses = sampledStatuses[len(sampledStatuses)-width:]
+		}
 	}
 
 	minL, maxL := samples[0], samples[0]
@@ -85,7 +89,7 @@ func latencySparkline(latencies []time.Duration, width int) string {
 		sb.WriteString(subtleStyle.Render(strings.Repeat("·", remaining)))
 	}
 	spread := maxL - minL
-	for _, l := range samples {
+	for i, l := range samples {
 		idx := 0
 		if spread > 0 {
 			idx = int(float64(l-minL) / float64(spread) * 7)
@@ -94,6 +98,10 @@ func latencySparkline(latencies []time.Duration, width int) string {
 			}
 		}
 		ch := string(sparkChars[idx])
+		isDown := i < len(sampledStatuses) && !sampledStatuses[i]
+		if isDown {
+			sb.WriteString(dangerStyle.Render(ch))
+		} else {
 			ms := l.Milliseconds()
 			if ms < 200 {
 				sb.WriteString(specialStyle.Render(ch))
@@ -103,6 +111,7 @@ func latencySparkline(latencies []time.Duration, width int) string {
 				sb.WriteString(dangerStyle.Render(ch))
 			}
 		}
+	}
 	return sb.String()
 }
 
@@ -474,7 +483,7 @@ func (m Model) viewSitesTab() string {
 				if site.Type == "push" {
 					spark = heartbeatSparkline(hist.Statuses, sparkWidth)
 				} else {
-					spark = latencySparkline(hist.Latencies, sparkWidth)
+					spark = latencySparkline(hist.Latencies, hist.Statuses, sparkWidth)
 				}
 
 				rows = append(rows, []string{
@@ -949,20 +958,27 @@ func (m Model) viewDetailPanel() string {
 				up, len(hist.Statuses))
 		}
 	} else {
-		b.WriteString("  " + latencySparkline(hist.Latencies, sparkWidth))
-		if len(hist.Latencies) > 0 {
-			minL, maxL := hist.Latencies[0], hist.Latencies[0]
-			var total time.Duration
-			for _, l := range hist.Latencies {
-				total += l
-				if l < minL {
-					minL = l
+		b.WriteString("  " + latencySparkline(hist.Latencies, hist.Statuses, sparkWidth))
+		// Stats over successful checks only — a failed check is stored as 0ns latency
+		// and would otherwise drag Min to 0ms and skew the average.
+		var minL, maxL, total time.Duration
+		count := 0
+		for i, l := range hist.Latencies {
+			if i < len(hist.Statuses) && !hist.Statuses[i] {
+				continue
 			}
-				if l > maxL {
+			if count == 0 {
+				minL, maxL = l, l
+			} else if l < minL {
+				minL = l
+			} else if l > maxL {
 				maxL = l
 			}
+			total += l
+			count++
 		}
-			avg := total / time.Duration(len(hist.Latencies))
+		if count > 0 {
+			avg := total / time.Duration(count)
 			fmt.Fprintf(&b, "\n  %s %dms  %s %dms  %s %dms",
 				subtleStyle.Render("Min"), minL.Milliseconds(),
 				subtleStyle.Render("Avg"), avg.Milliseconds(),

internal/tui/tui.go

@@ -4,6 +4,7 @@ import (
 	"encoding/json"
 	"fmt"
 	"math"
+	"os"
 	"sort"
 	"strings"
 	"time"
@@ -122,6 +123,10 @@ type Model struct {
 
 	filterMode bool
 	filterText string
+
+	// demoMode renders a stable status dot instead of the animated pulse so
+	// screenshots/recordings don't capture the spinner mid-frame. Set via UPTOP_DEMO=1.
+	demoMode bool
 }
 
 func InitialModel(isAdmin bool, s store.Store, eng *monitor.Engine) Model {
@@ -155,6 +160,7 @@ func InitialModel(isAdmin bool, s store.Store, eng *monitor.Engine) Model {
 		collapsed:    collapsed,
 		theme:        theme,
 		themeIndex:   themeIdx,
+		demoMode:     os.Getenv("UPTOP_DEMO") == "1",
 	}
 }
 
@@ -754,11 +760,6 @@ func (m *Model) submitForm() {
 }
 
 func (m Model) pulseIndicator() string {
-	frame := m.tickCount % len(pulseFrames)
-	brightness := int(m.pulsePos*155) + 100
-	if brightness > 255 {
-		brightness = 255
-	}
 	hasDown := false
 	for _, s := range m.sites {
 		if !s.Paused && !m.isMonitorInMaintenance(s.ID) && (s.Status == "DOWN" || s.Status == "SSL EXP") {
@@ -766,6 +767,19 @@ func (m Model) pulseIndicator() string {
 			break
 		}
 	}
+	// Stills can't show animation: render a stable status dot in demo mode.
+	if m.demoMode {
+		c := m.theme.Success
+		if hasDown {
+			c = m.theme.Danger
+		}
+		return lipgloss.NewStyle().Foreground(c).Render("●")
+	}
+	frame := m.tickCount % len(pulseFrames)
+	brightness := int(m.pulsePos*155) + 100
+	if brightness > 255 {
+		brightness = 255
+	}
 	var color string
 	if hasDown {
 		color = fmt.Sprintf("#%02x%02x%02x", brightness, brightness/4, brightness/4)
@@ -953,7 +967,11 @@ func (m Model) viewDashboard() string {
 				online++
 			}
 		}
-		statusParts = append(statusParts, fmt.Sprintf("%d probes", online))
+		probeLabel := "probes"
+		if online == 1 {
+			probeLabel = "probe"
+		}
+		statusParts = append(statusParts, fmt.Sprintf("%d %s", online, probeLabel))
 	}
 	statusLine := strings.Join(statusParts, subtleStyle.Render(" · "))