feat(bot): report Telegram bot Bot API health to the gateway over the bot-link

The bot runs on its own host and exports no telemetry (otelcol is unreachable
from there), so it was a monitoring blind spot. Observe its Bot API health
centrally by wrapping the HTTP client — one place, no per-call-site
instrumentation — and relay it up the existing bot-link as a periodic Health
message the gateway turns into its own metrics.

- proto: add Health (delta connect / api / 429 counters + a last-ok stamp) to
  the FromBot oneof (additive, backward-compatible).
- bot: platform/telegram/internal/health wraps the Bot API HTTP client — it
  stamps liveness on any 2xx (so the getUpdates long-poll keeps it fresh even
  when idle), classifies transport/5xx failures (getUpdates vs other) and 429s,
  and honours a 429's Retry-After (bounded) so the bot backs off; a 4xx other
  than 429 is a normal per-request outcome and is not counted.
- bot-link client: flush the reporter as a Health message every 30s over a
  single-sender loop (a gRPC stream forbids concurrent Send).
- gateway: fold each report into bot_tg_errors_total{kind} and the
  bot_tg_last_ok_unix liveness gauge.
- grafana: alerts (bot disconnected, bot not reaching the Bot API, sustained
  429s) routed to the operator email — which does not go through the bot — plus
  a Telegram-bot dashboard.
- docs (ARCHITECTURE, compose comments); unit tests (observer classification,
  Retry-After, snapshot/commit, hub last-ok monotonicity).

deploy/docker-compose.bot.yml

@@ -3,8 +3,10 @@
 #   docker compose -f docker-compose.bot.yml up -d
 #
 # The bot egresses to the Bot API directly (no VPN sidecar) and dials the main host's
-# published bot-link :9443 over mTLS. It exports no telemetry — otelcol lives on the
-# main host and is unreachable from here — so observe it via `docker logs` on this host.
+# published bot-link :9443 over mTLS. It exports no OTLP telemetry — otelcol lives on the
+# main host and is unreachable from here — but it reports its Bot API health up the bot-link,
+# which the gateway turns into metrics + alerts on the main host (docs/ARCHITECTURE.md); `docker
+# logs` on this host is the local detail view.
 # Values come from the prod-deploy workflow (PROD_ secrets/variables); BOT_IMAGE is the
 # pushed registry tag and BOTLINK_GATEWAY_ADDR is the main host's <ip>:9443.
 name: scrabble-bot
@@ -50,7 +52,8 @@ services:
       TELEGRAM_BOTLINK_TLS_CA: /certs/ca.crt
       TELEGRAM_LOG_LEVEL: ${LOG_LEVEL:-info}
       TELEGRAM_SERVICE_NAME: scrabble-telegram-bot
-      # No telemetry export: otelcol is on the main host, unreachable from here.
+      # No OTLP export: otelcol is on the main host, unreachable from here. Bot API health rides the
+      # bot-link instead (the gateway exposes it as metrics); see docs/ARCHITECTURE.md.
       TELEGRAM_OTEL_TRACES_EXPORTER: none
       TELEGRAM_OTEL_METRICS_EXPORTER: none
       GOMAXPROCS: "1"

deploy/grafana/dashboards/bot.json

@@ -0,0 +1,45 @@
+{
+  "uid": "scrabble-bot",
+  "title": "Scrabble — Telegram bot",
+  "tags": ["scrabble"],
+  "timezone": "",
+  "schemaVersion": 39,
+  "version": 1,
+  "refresh": "30s",
+  "time": { "from": "now-6h", "to": "now" },
+  "panels": [
+    {
+      "type": "stat",
+      "title": "Bot connected",
+      "description": "botlink_connected_bots: bots currently holding the gateway bot-link (1 = healthy).",
+      "gridPos": { "h": 5, "w": 6, "x": 0, "y": 0 },
+      "datasource": { "type": "prometheus", "uid": "prometheus" },
+      "targets": [{ "refId": "A", "expr": "max(botlink_connected_bots)" }]
+    },
+    {
+      "type": "stat",
+      "title": "Last Bot API OK (age)",
+      "description": "Seconds since the bot's most recent successful Bot API call. Grows unbounded if the bot wedges.",
+      "gridPos": { "h": 5, "w": 6, "x": 6, "y": 0 },
+      "fieldConfig": { "defaults": { "unit": "s" }, "overrides": [] },
+      "datasource": { "type": "prometheus", "uid": "prometheus" },
+      "targets": [{ "refId": "A", "expr": "time() - max(bot_tg_last_ok_unix)" }]
+    },
+    {
+      "type": "timeseries",
+      "title": "Bot API errors/s by kind",
+      "description": "bot_tg_errors_total by kind: connect (getUpdates), api (other sends), rate_limited (429). 429 should stay ~0.",
+      "gridPos": { "h": 8, "w": 12, "x": 12, "y": 0 },
+      "datasource": { "type": "prometheus", "uid": "prometheus" },
+      "targets": [{ "refId": "A", "expr": "sum by (kind) (rate(bot_tg_errors_total[5m]))", "legendFormat": "{{kind}}" }]
+    },
+    {
+      "type": "timeseries",
+      "title": "Bot-link commands/s by result",
+      "description": "botlink_commands_total by result: delivered / not_delivered / dropped / error. Sustained not_delivered or error means gateway sends are failing to reach Telegram.",
+      "gridPos": { "h": 8, "w": 12, "x": 0, "y": 5 },
+      "datasource": { "type": "prometheus", "uid": "prometheus" },
+      "targets": [{ "refId": "A", "expr": "sum by (result) (rate(botlink_commands_total[5m]))", "legendFormat": "{{result}}" }]
+    }
+  ]
+}

deploy/grafana/provisioning/alerting/rules.yaml

@@ -277,3 +277,91 @@ groups:
                 - evaluator: { type: gt, params: [1800] }
         labels: { severity: critical }
         annotations: { summary: 'No WAL archived for over 30 minutes while pg_wal keeps growing — archiving is genuinely stalled; the PITR recovery point is frozen and pg_wal will fill the disk. Run `docker exec scrabble-postgres pgbackrest --stanza=scrabble --pg1-user=scrabble check`.' }
+
+  # Remote Telegram bot health. The bot runs on its own host and exports no telemetry of its own; the
+  # gateway observes it through the bot-link (botlink_connected_bots) and the health it reports over
+  # that stream (bot_tg_*). All absent/NaN-safe: before a bot ever connects the metrics are absent, so
+  # noDataState OK keeps them quiet on a fresh contour. The alert email does NOT go through the bot, so
+  # a bot-down alert is deliverable.
+  - orgId: 1
+    name: scrabble-bot
+    folder: Alerts
+    interval: 1m
+    rules:
+      - uid: bot_disconnected
+        title: Telegram bot disconnected
+        condition: C
+        for: 5m
+        noDataState: OK
+        execErrState: OK
+        data:
+          - refId: A
+            relativeTimeRange: { from: 300, to: 0 }
+            datasourceUid: prometheus
+            model: { refId: A, expr: botlink_connected_bots, instant: true }
+          - refId: C
+            datasourceUid: __expr__
+            model:
+              refId: C
+              type: threshold
+              expression: A
+              conditions:
+                - evaluator: { type: lt, params: [1] }
+        labels: { severity: critical }
+        annotations: { summary: 'No Telegram bot is connected to the gateway bot-link — out-of-app push and admin sends are down. Check the bot host.' }
+
+      # Positive liveness: a bot is connected but its last successful Bot API call is over 5 minutes
+      # old — the getUpdates long-poll returns every ~minute even when idle, so a stale stamp means the
+      # bot is wedged (no errors, no traffic). Guarded by "and connected" so a mere disconnect (owned by
+      # bot_disconnected) does not double-fire.
+      - uid: bot_tg_stale
+        title: Telegram bot not reaching the Bot API
+        condition: C
+        for: 5m
+        noDataState: OK
+        execErrState: OK
+        data:
+          - refId: A
+            relativeTimeRange: { from: 600, to: 0 }
+            datasourceUid: prometheus
+            model:
+              refId: A
+              expr: (time() - bot_tg_last_ok_unix) and on() (botlink_connected_bots >= 1)
+              instant: true
+          - refId: C
+            datasourceUid: __expr__
+            model:
+              refId: C
+              type: threshold
+              expression: A
+              conditions:
+                - evaluator: { type: gt, params: [300] }
+        labels: { severity: critical }
+        annotations: { summary: 'A connected Telegram bot has not reached the Bot API in over 5 minutes — the update poll is wedged. Check the bot host and Telegram reachability.' }
+
+      # 429s should be ~never once the bot honours Retry-After; any sustained rate-limiting is a symptom
+      # (a send loop, a misbehaving path) worth investigating.
+      - uid: bot_tg_rate_limited
+        title: Telegram bot rate-limited (429)
+        condition: C
+        for: 5m
+        noDataState: OK
+        execErrState: OK
+        data:
+          - refId: A
+            relativeTimeRange: { from: 900, to: 0 }
+            datasourceUid: prometheus
+            model:
+              refId: A
+              expr: increase(bot_tg_errors_total{kind="rate_limited"}[15m])
+              instant: true
+          - refId: C
+            datasourceUid: __expr__
+            model:
+              refId: C
+              type: threshold
+              expression: A
+              conditions:
+                - evaluator: { type: gt, params: [0] }
+        labels: { severity: warning }
+        annotations: { summary: 'The Telegram bot is being rate-limited (HTTP 429). It should honour Retry-After, so sustained 429s point to a send loop or a hot path.' }

docs/ARCHITECTURE.md

@@ -1015,6 +1015,20 @@ answering `/start` with a URL button into the **main** bot's Mini App (`?startap
 button would sign initData with the promo token); it is self-contained — no bot-link, no gateway.
 Session-revocation events and cursor-based stream resume stay deferred (single-instance MVP).
 
+Because the bot **exports no telemetry of its own** (the OTel collector is on the main host,
+unreachable from the bot host), it reports its **Bot API health** up the same stream: a periodic
+`Health` message (`platform/telegram/internal/health`) carries delta counts of connect failures (the
+getUpdates long-poll), other API errors and 429s, plus the wall-clock second of its last successful
+Bot API call. The bot observes these **centrally by wrapping the Bot API HTTP client** — one place,
+no per-call-site instrumentation — and there also honours a 429's `Retry-After` (bounded) so it backs
+off rather than hammering (a 429 should therefore stay ~0). The gateway folds each report into its
+own metrics (`bot_tg_errors_total{kind}`, the `bot_tg_last_ok_unix` liveness gauge). The remote bot
+is thus monitored from the main host's Grafana with three layered signals: the **connection gauge**
+(`botlink_connected_bots`) catches a link/host/process outage, the **liveness gauge** catches a
+silently wedged bot (its stamp stays fresh even when idle, since getUpdates returns every poll), and
+**`botlink_commands_total{result}`** catches gateway sends failing to reach Telegram. Alerts route to
+the operator email, which does not depend on the bot.
+
 A separate **advertising-banner** channel feeds the client's one-line strip (UI_DESIGN.md),
 server-driven by `internal/ads`. An operator manages **campaigns** (each one placement order) in
 the admin console (`/_gm/banners`): a campaign has a show **weight** (integer percent 1..100), an
@@ -1073,7 +1087,9 @@ link — misses the event; while an add-email confirmation is pending the client
   `OTEL_EXPORTER_OTLP_*` environment) exports to a collector. The Postgres pool is
   instrumented with otelsql and `otelgrpc` traces the backend↔gateway push stream
   and the gateway↔validator and bot-link calls; the gateway also exports
-  `botlink_connected_bots` and `botlink_commands_total` (by result) for the bot-link. The OTLP **Collector** (OTLP/gRPC → Prometheus
+  `botlink_connected_bots` and `botlink_commands_total` (by result) for the bot-link, plus the remote
+  bot's own Bot API health it relays over the stream — `bot_tg_errors_total` (by kind) and the
+  `bot_tg_last_ok_unix` liveness gauge (see the bot-link section). The OTLP **Collector** (OTLP/gRPC → Prometheus
   metrics + Tempo traces), **Prometheus** (15d), **Tempo** (72h) and **Grafana**
   (provisioned datasources + dashboards, behind the caddy `/_gm/grafana` Basic-Auth)
   are stood up with the deploy (`deploy/`); the default exporter stays

gateway/internal/botlink/hub.go

@@ -76,6 +76,11 @@ type Hub struct {
 
 	connected metric.Int64UpDownCounter
 	commands  metric.Int64Counter
+	// tgErrors counts the remote bot's Bot API failures it reports over the stream, by kind
+	// (connect / api / rate_limited). lastOK holds the wall-clock second of the bot's most recent
+	// successful Bot API call, read by the bot_tg_last_ok_unix observable gauge for a staleness alert.
+	tgErrors metric.Int64Counter
+	lastOK   atomic.Int64
 }
 
 // link is one connected bot's outbound queue and identity.
@@ -103,6 +108,19 @@ func NewHub(log *zap.Logger, meter metric.Meter, resolve EligibilityResolver) *H
 			metric.WithDescription("Number of Telegram bots currently connected to the gateway bot-link."))
 		h.commands, _ = meter.Int64Counter("botlink_commands_total",
 			metric.WithDescription("Bot-link send commands by result (delivered, not_delivered, dropped, error)."))
+		h.tgErrors, _ = meter.Int64Counter("bot_tg_errors_total",
+			metric.WithDescription("Telegram Bot API failures the remote bot reports over the bot-link, by kind (connect, api, rate_limited)."))
+		// The bot's last successful Bot API second, reported over the stream. An observable gauge
+		// reads the atomic on each collection; it observes nothing until a bot has reported, so a
+		// never-connected gateway shows no data (the connected-bots alert covers that case).
+		_, _ = meter.Int64ObservableGauge("bot_tg_last_ok_unix",
+			metric.WithDescription("Unix second of the remote bot's most recent successful Bot API call (0/absent until reported)."),
+			metric.WithInt64Callback(func(_ context.Context, o metric.Int64Observer) error {
+				if v := h.lastOK.Load(); v > 0 {
+					o.Observe(v)
+				}
+				return nil
+			}))
 	}
 	return h
 }
@@ -149,6 +167,36 @@ func (h *Hub) Link(stream grpc.BidiStreamingServer[botlinkv1.FromBot, botlinkv1.
 		if ack := msg.GetAck(); ack != nil {
 			h.resolve(ack)
 		}
+		if hb := msg.GetHealth(); hb != nil {
+			h.recordHealth(hb)
+		}
+	}
+}
+
+// recordHealth folds one bot Health report into the gateway's cumulative Bot API metrics: the three
+// delta counters are added under their kind label, and the last-ok stamp (an absolute wall-clock
+// second) advances the liveness gauge (monotonically — a stale reordered report cannot rewind it).
+func (h *Hub) recordHealth(hb *botlinkv1.Health) {
+	if h.tgErrors != nil {
+		ctx := context.Background()
+		if v := hb.GetConnectFailures(); v > 0 {
+			h.tgErrors.Add(ctx, int64(v), metric.WithAttributes(attribute.String("kind", "connect")))
+		}
+		if v := hb.GetApiErrors(); v > 0 {
+			h.tgErrors.Add(ctx, int64(v), metric.WithAttributes(attribute.String("kind", "api")))
+		}
+		if v := hb.GetRateLimited(); v > 0 {
+			h.tgErrors.Add(ctx, int64(v), metric.WithAttributes(attribute.String("kind", "rate_limited")))
+		}
+	}
+	for {
+		cur := h.lastOK.Load()
+		if hb.GetLastOkUnix() <= cur {
+			break
+		}
+		if h.lastOK.CompareAndSwap(cur, hb.GetLastOkUnix()) {
+			break
+		}
 	}
 }
 

gateway/internal/botlink/hub_test.go

@@ -233,3 +233,21 @@ func TestRelayServerNoBot(t *testing.T) {
 		t.Fatal("expected an error with no bot connected")
 	}
 }
+
+// TestRecordHealthLastOKMonotonic checks a bot Health report advances the last-ok liveness stamp but
+// a stale or reordered report (an earlier second) never rewinds it.
+func TestRecordHealthLastOKMonotonic(t *testing.T) {
+	hub := NewHub(nil, nil, nil)
+	hub.recordHealth(&botlinkv1.Health{LastOkUnix: 100})
+	if got := hub.lastOK.Load(); got != 100 {
+		t.Fatalf("lastOK = %d, want 100", got)
+	}
+	hub.recordHealth(&botlinkv1.Health{LastOkUnix: 90}) // reordered/stale — must not rewind
+	if got := hub.lastOK.Load(); got != 100 {
+		t.Errorf("lastOK rewound to %d, want 100", got)
+	}
+	hub.recordHealth(&botlinkv1.Health{LastOkUnix: 150})
+	if got := hub.lastOK.Load(); got != 150 {
+		t.Errorf("lastOK = %d, want 150", got)
+	}
+}

pkg/proto/botlink/v1/botlink.pb.go

@@ -30,13 +30,14 @@ const (
 )
 
 // FromBot is a message the bot sends to the gateway: the opening Hello, then one
-// Ack per Command.
+// Ack per Command and a periodic Health report.
 type FromBot struct {
 	state protoimpl.MessageState `protogen:"open.v1"`
 	// Types that are valid to be assigned to Msg:
 	//
 	//	*FromBot_Hello
 	//	*FromBot_Ack
+	//	*FromBot_Health
 	Msg           isFromBot_Msg `protobuf_oneof:"msg"`
 	unknownFields protoimpl.UnknownFields
 	sizeCache     protoimpl.SizeCache
@@ -97,6 +98,15 @@ func (x *FromBot) GetAck() *Ack {
 	return nil
 }
 
+func (x *FromBot) GetHealth() *Health {
+	if x != nil {
+		if x, ok := x.Msg.(*FromBot_Health); ok {
+			return x.Health
+		}
+	}
+	return nil
+}
+
 type isFromBot_Msg interface {
 	isFromBot_Msg()
 }
@@ -109,10 +119,92 @@ type FromBot_Ack struct {
 	Ack *Ack `protobuf:"bytes,2,opt,name=ack,proto3,oneof"`
 }
 
+type FromBot_Health struct {
+	Health *Health `protobuf:"bytes,3,opt,name=health,proto3,oneof"`
+}
+
 func (*FromBot_Hello) isFromBot_Msg() {}
 
 func (*FromBot_Ack) isFromBot_Msg() {}
 
+func (*FromBot_Health) isFromBot_Msg() {}
+
+// Health is a periodic report the bot pushes up the stream so the gateway can expose the remote
+// bot's Bot-API health as its own metrics — the bot exports no telemetry of its own (otelcol lives
+// on the main host, unreachable from the bot), so the bot-link is its only channel out. The three
+// counters are DELTAS since the previous Health: the gateway adds them to cumulative counters, so a
+// report lost across a reconnect only undercounts, never corrupts. last_ok_unix is the wall-clock
+// second of the bot's most recent successful Bot API response (any call, including the getUpdates
+// long-poll that returns every poll cycle even when idle) — an absolute liveness stamp the gateway
+// surfaces as a gauge, so a silently stalled bot (no errors, no traffic) is still caught.
+type Health struct {
+	state           protoimpl.MessageState `protogen:"open.v1"`
+	ConnectFailures uint64                 `protobuf:"varint,1,opt,name=connect_failures,json=connectFailures,proto3" json:"connect_failures,omitempty"` // getUpdates long-poll failures (transport / 5xx) since the last report
+	ApiErrors       uint64                 `protobuf:"varint,2,opt,name=api_errors,json=apiErrors,proto3" json:"api_errors,omitempty"`                   // other Bot API failures (transport / 5xx) since the last report
+	RateLimited     uint64                 `protobuf:"varint,3,opt,name=rate_limited,json=rateLimited,proto3" json:"rate_limited,omitempty"`             // Bot API 429 responses since the last report (should stay ~0)
+	LastOkUnix      int64                  `protobuf:"varint,4,opt,name=last_ok_unix,json=lastOkUnix,proto3" json:"last_ok_unix,omitempty"`              // unix seconds of the last successful Bot API response (0 = none yet)
+	unknownFields   protoimpl.UnknownFields
+	sizeCache       protoimpl.SizeCache
+}
+
+func (x *Health) Reset() {
+	*x = Health{}
+	mi := &file_botlink_v1_botlink_proto_msgTypes[1]
+	ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
+	ms.StoreMessageInfo(mi)
+}
+
+func (x *Health) String() string {
+	return protoimpl.X.MessageStringOf(x)
+}
+
+func (*Health) ProtoMessage() {}
+
+func (x *Health) ProtoReflect() protoreflect.Message {
+	mi := &file_botlink_v1_botlink_proto_msgTypes[1]
+	if x != nil {
+		ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
+		if ms.LoadMessageInfo() == nil {
+			ms.StoreMessageInfo(mi)
+		}
+		return ms
+	}
+	return mi.MessageOf(x)
+}
+
+// Deprecated: Use Health.ProtoReflect.Descriptor instead.
+func (*Health) Descriptor() ([]byte, []int) {
+	return file_botlink_v1_botlink_proto_rawDescGZIP(), []int{1}
+}
+
+func (x *Health) GetConnectFailures() uint64 {
+	if x != nil {
+		return x.ConnectFailures
+	}
+	return 0
+}
+
+func (x *Health) GetApiErrors() uint64 {
+	if x != nil {
+		return x.ApiErrors
+	}
+	return 0
+}
+
+func (x *Health) GetRateLimited() uint64 {
+	if x != nil {
+		return x.RateLimited
+	}
+	return 0
+}
+
+func (x *Health) GetLastOkUnix() int64 {
+	if x != nil {
+		return x.LastOkUnix
+	}
+	return 0
+}
+
 // ToBot is a message the gateway sends to the bot. Only Command is carried today.
 type ToBot struct {
 	state         protoimpl.MessageState `protogen:"open.v1"`
@@ -123,7 +215,7 @@ type ToBot struct {
 
 func (x *ToBot) Reset() {
 	*x = ToBot{}
-	mi := &file_botlink_v1_botlink_proto_msgTypes[1]
+	mi := &file_botlink_v1_botlink_proto_msgTypes[2]
 	ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
 	ms.StoreMessageInfo(mi)
 }
@@ -135,7 +227,7 @@ func (x *ToBot) String() string {
 func (*ToBot) ProtoMessage() {}
 
 func (x *ToBot) ProtoReflect() protoreflect.Message {
-	mi := &file_botlink_v1_botlink_proto_msgTypes[1]
+	mi := &file_botlink_v1_botlink_proto_msgTypes[2]
 	if x != nil {
 		ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
 		if ms.LoadMessageInfo() == nil {
@@ -148,7 +240,7 @@ func (x *ToBot) ProtoReflect() protoreflect.Message {
 
 // Deprecated: Use ToBot.ProtoReflect.Descriptor instead.
 func (*ToBot) Descriptor() ([]byte, []int) {
-	return file_botlink_v1_botlink_proto_rawDescGZIP(), []int{1}
+	return file_botlink_v1_botlink_proto_rawDescGZIP(), []int{2}
 }
 
 func (x *ToBot) GetCommand() *Command {
@@ -171,7 +263,7 @@ type Hello struct {
 
 func (x *Hello) Reset() {
 	*x = Hello{}
-	mi := &file_botlink_v1_botlink_proto_msgTypes[2]
+	mi := &file_botlink_v1_botlink_proto_msgTypes[3]
 	ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
 	ms.StoreMessageInfo(mi)
 }
@@ -183,7 +275,7 @@ func (x *Hello) String() string {
 func (*Hello) ProtoMessage() {}
 
 func (x *Hello) ProtoReflect() protoreflect.Message {
-	mi := &file_botlink_v1_botlink_proto_msgTypes[2]
+	mi := &file_botlink_v1_botlink_proto_msgTypes[3]
 	if x != nil {
 		ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
 		if ms.LoadMessageInfo() == nil {
@@ -196,7 +288,7 @@ func (x *Hello) ProtoReflect() protoreflect.Message {
 
 // Deprecated: Use Hello.ProtoReflect.Descriptor instead.
 func (*Hello) Descriptor() ([]byte, []int) {
-	return file_botlink_v1_botlink_proto_rawDescGZIP(), []int{2}
+	return file_botlink_v1_botlink_proto_rawDescGZIP(), []int{3}
 }
 
 func (x *Hello) GetInstanceId() string {
@@ -233,7 +325,7 @@ type Command struct {
 
 func (x *Command) Reset() {
 	*x = Command{}
-	mi := &file_botlink_v1_botlink_proto_msgTypes[3]
+	mi := &file_botlink_v1_botlink_proto_msgTypes[4]
 	ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
 	ms.StoreMessageInfo(mi)
 }
@@ -245,7 +337,7 @@ func (x *Command) String() string {
 func (*Command) ProtoMessage() {}
 
 func (x *Command) ProtoReflect() protoreflect.Message {
-	mi := &file_botlink_v1_botlink_proto_msgTypes[3]
+	mi := &file_botlink_v1_botlink_proto_msgTypes[4]
 	if x != nil {
 		ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
 		if ms.LoadMessageInfo() == nil {
@@ -258,7 +350,7 @@ func (x *Command) ProtoReflect() protoreflect.Message {
 
 // Deprecated: Use Command.ProtoReflect.Descriptor instead.
 func (*Command) Descriptor() ([]byte, []int) {
-	return file_botlink_v1_botlink_proto_rawDescGZIP(), []int{3}
+	return file_botlink_v1_botlink_proto_rawDescGZIP(), []int{4}
 }
 
 func (x *Command) GetCommandId() string {
@@ -372,7 +464,7 @@ type Ack struct {
 
 func (x *Ack) Reset() {
 	*x = Ack{}
-	mi := &file_botlink_v1_botlink_proto_msgTypes[4]
+	mi := &file_botlink_v1_botlink_proto_msgTypes[5]
 	ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
 	ms.StoreMessageInfo(mi)
 }
@@ -384,7 +476,7 @@ func (x *Ack) String() string {
 func (*Ack) ProtoMessage() {}
 
 func (x *Ack) ProtoReflect() protoreflect.Message {
-	mi := &file_botlink_v1_botlink_proto_msgTypes[4]
+	mi := &file_botlink_v1_botlink_proto_msgTypes[5]
 	if x != nil {
 		ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
 		if ms.LoadMessageInfo() == nil {
@@ -397,7 +489,7 @@ func (x *Ack) ProtoReflect() protoreflect.Message {
 
 // Deprecated: Use Ack.ProtoReflect.Descriptor instead.
 func (*Ack) Descriptor() ([]byte, []int) {
-	return file_botlink_v1_botlink_proto_rawDescGZIP(), []int{4}
+	return file_botlink_v1_botlink_proto_rawDescGZIP(), []int{5}
 }
 
 func (x *Ack) GetCommandId() string {
@@ -445,7 +537,7 @@ type ChatGateCommand struct {
 
 func (x *ChatGateCommand) Reset() {
 	*x = ChatGateCommand{}
-	mi := &file_botlink_v1_botlink_proto_msgTypes[5]
+	mi := &file_botlink_v1_botlink_proto_msgTypes[6]
 	ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
 	ms.StoreMessageInfo(mi)
 }
@@ -457,7 +549,7 @@ func (x *ChatGateCommand) String() string {
 func (*ChatGateCommand) ProtoMessage() {}
 
 func (x *ChatGateCommand) ProtoReflect() protoreflect.Message {
-	mi := &file_botlink_v1_botlink_proto_msgTypes[5]
+	mi := &file_botlink_v1_botlink_proto_msgTypes[6]
 	if x != nil {
 		ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
 		if ms.LoadMessageInfo() == nil {
@@ -470,7 +562,7 @@ func (x *ChatGateCommand) ProtoReflect() protoreflect.Message {
 
 // Deprecated: Use ChatGateCommand.ProtoReflect.Descriptor instead.
 func (*ChatGateCommand) Descriptor() ([]byte, []int) {
-	return file_botlink_v1_botlink_proto_rawDescGZIP(), []int{5}
+	return file_botlink_v1_botlink_proto_rawDescGZIP(), []int{6}
 }
 
 func (x *ChatGateCommand) GetExternalId() string {
@@ -498,7 +590,7 @@ type ChatEligibilityRequest struct {
 
 func (x *ChatEligibilityRequest) Reset() {
 	*x = ChatEligibilityRequest{}
-	mi := &file_botlink_v1_botlink_proto_msgTypes[6]
+	mi := &file_botlink_v1_botlink_proto_msgTypes[7]
 	ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
 	ms.StoreMessageInfo(mi)
 }
@@ -510,7 +602,7 @@ func (x *ChatEligibilityRequest) String() string {
 func (*ChatEligibilityRequest) ProtoMessage() {}
 
 func (x *ChatEligibilityRequest) ProtoReflect() protoreflect.Message {
-	mi := &file_botlink_v1_botlink_proto_msgTypes[6]
+	mi := &file_botlink_v1_botlink_proto_msgTypes[7]
 	if x != nil {
 		ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
 		if ms.LoadMessageInfo() == nil {
@@ -523,7 +615,7 @@ func (x *ChatEligibilityRequest) ProtoReflect() protoreflect.Message {
 
 // Deprecated: Use ChatEligibilityRequest.ProtoReflect.Descriptor instead.
 func (*ChatEligibilityRequest) Descriptor() ([]byte, []int) {
-	return file_botlink_v1_botlink_proto_rawDescGZIP(), []int{6}
+	return file_botlink_v1_botlink_proto_rawDescGZIP(), []int{7}
 }
 
 func (x *ChatEligibilityRequest) GetExternalId() string {
@@ -546,7 +638,7 @@ type ChatEligibilityResponse struct {
 
 func (x *ChatEligibilityResponse) Reset() {
 	*x = ChatEligibilityResponse{}
-	mi := &file_botlink_v1_botlink_proto_msgTypes[7]
+	mi := &file_botlink_v1_botlink_proto_msgTypes[8]
 	ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
 	ms.StoreMessageInfo(mi)
 }
@@ -558,7 +650,7 @@ func (x *ChatEligibilityResponse) String() string {
 func (*ChatEligibilityResponse) ProtoMessage() {}
 
 func (x *ChatEligibilityResponse) ProtoReflect() protoreflect.Message {
-	mi := &file_botlink_v1_botlink_proto_msgTypes[7]
+	mi := &file_botlink_v1_botlink_proto_msgTypes[8]
 	if x != nil {
 		ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
 		if ms.LoadMessageInfo() == nil {
@@ -571,7 +663,7 @@ func (x *ChatEligibilityResponse) ProtoReflect() protoreflect.Message {
 
 // Deprecated: Use ChatEligibilityResponse.ProtoReflect.Descriptor instead.
 func (*ChatEligibilityResponse) Descriptor() ([]byte, []int) {
-	return file_botlink_v1_botlink_proto_rawDescGZIP(), []int{7}
+	return file_botlink_v1_botlink_proto_rawDescGZIP(), []int{8}
 }
 
 func (x *ChatEligibilityResponse) GetRegistered() bool {
@@ -605,7 +697,7 @@ type CreateInvoiceCommand struct {
 
 func (x *CreateInvoiceCommand) Reset() {
 	*x = CreateInvoiceCommand{}
-	mi := &file_botlink_v1_botlink_proto_msgTypes[8]
+	mi := &file_botlink_v1_botlink_proto_msgTypes[9]
 	ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
 	ms.StoreMessageInfo(mi)
 }
@@ -617,7 +709,7 @@ func (x *CreateInvoiceCommand) String() string {
 func (*CreateInvoiceCommand) ProtoMessage() {}
 
 func (x *CreateInvoiceCommand) ProtoReflect() protoreflect.Message {
-	mi := &file_botlink_v1_botlink_proto_msgTypes[8]
+	mi := &file_botlink_v1_botlink_proto_msgTypes[9]
 	if x != nil {
 		ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
 		if ms.LoadMessageInfo() == nil {
@@ -630,7 +722,7 @@ func (x *CreateInvoiceCommand) ProtoReflect() protoreflect.Message {
 
 // Deprecated: Use CreateInvoiceCommand.ProtoReflect.Descriptor instead.
 func (*CreateInvoiceCommand) Descriptor() ([]byte, []int) {
-	return file_botlink_v1_botlink_proto_rawDescGZIP(), []int{8}
+	return file_botlink_v1_botlink_proto_rawDescGZIP(), []int{9}
 }
 
 func (x *CreateInvoiceCommand) GetTitle() string {
@@ -674,7 +766,7 @@ type PreCheckoutRequest struct {
 
 func (x *PreCheckoutRequest) Reset() {
 	*x = PreCheckoutRequest{}
-	mi := &file_botlink_v1_botlink_proto_msgTypes[9]
+	mi := &file_botlink_v1_botlink_proto_msgTypes[10]
 	ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
 	ms.StoreMessageInfo(mi)
 }
@@ -686,7 +778,7 @@ func (x *PreCheckoutRequest) String() string {
 func (*PreCheckoutRequest) ProtoMessage() {}
 
 func (x *PreCheckoutRequest) ProtoReflect() protoreflect.Message {
-	mi := &file_botlink_v1_botlink_proto_msgTypes[9]
+	mi := &file_botlink_v1_botlink_proto_msgTypes[10]
 	if x != nil {
 		ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
 		if ms.LoadMessageInfo() == nil {
@@ -699,7 +791,7 @@ func (x *PreCheckoutRequest) ProtoReflect() protoreflect.Message {
 
 // Deprecated: Use PreCheckoutRequest.ProtoReflect.Descriptor instead.
 func (*PreCheckoutRequest) Descriptor() ([]byte, []int) {
-	return file_botlink_v1_botlink_proto_rawDescGZIP(), []int{9}
+	return file_botlink_v1_botlink_proto_rawDescGZIP(), []int{10}
 }
 
 func (x *PreCheckoutRequest) GetOrderId() string {
@@ -735,7 +827,7 @@ type PreCheckoutResponse struct {
 
 func (x *PreCheckoutResponse) Reset() {
 	*x = PreCheckoutResponse{}
-	mi := &file_botlink_v1_botlink_proto_msgTypes[10]
+	mi := &file_botlink_v1_botlink_proto_msgTypes[11]
 	ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
 	ms.StoreMessageInfo(mi)
 }
@@ -747,7 +839,7 @@ func (x *PreCheckoutResponse) String() string {
 func (*PreCheckoutResponse) ProtoMessage() {}
 
 func (x *PreCheckoutResponse) ProtoReflect() protoreflect.Message {
-	mi := &file_botlink_v1_botlink_proto_msgTypes[10]
+	mi := &file_botlink_v1_botlink_proto_msgTypes[11]
 	if x != nil {
 		ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
 		if ms.LoadMessageInfo() == nil {
@@ -760,7 +852,7 @@ func (x *PreCheckoutResponse) ProtoReflect() protoreflect.Message {
 
 // Deprecated: Use PreCheckoutResponse.ProtoReflect.Descriptor instead.
 func (*PreCheckoutResponse) Descriptor() ([]byte, []int) {
-	return file_botlink_v1_botlink_proto_rawDescGZIP(), []int{10}
+	return file_botlink_v1_botlink_proto_rawDescGZIP(), []int{11}
 }
 
 func (x *PreCheckoutResponse) GetOk() bool {
@@ -792,7 +884,7 @@ type ForwardPaymentRequest struct {
 
 func (x *ForwardPaymentRequest) Reset() {
 	*x = ForwardPaymentRequest{}
-	mi := &file_botlink_v1_botlink_proto_msgTypes[11]
+	mi := &file_botlink_v1_botlink_proto_msgTypes[12]
 	ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
 	ms.StoreMessageInfo(mi)
 }
@@ -804,7 +896,7 @@ func (x *ForwardPaymentRequest) String() string {
 func (*ForwardPaymentRequest) ProtoMessage() {}
 
 func (x *ForwardPaymentRequest) ProtoReflect() protoreflect.Message {
-	mi := &file_botlink_v1_botlink_proto_msgTypes[11]
+	mi := &file_botlink_v1_botlink_proto_msgTypes[12]
 	if x != nil {
 		ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
 		if ms.LoadMessageInfo() == nil {
@@ -817,7 +909,7 @@ func (x *ForwardPaymentRequest) ProtoReflect() protoreflect.Message {
 
 // Deprecated: Use ForwardPaymentRequest.ProtoReflect.Descriptor instead.
 func (*ForwardPaymentRequest) Descriptor() ([]byte, []int) {
-	return file_botlink_v1_botlink_proto_rawDescGZIP(), []int{11}
+	return file_botlink_v1_botlink_proto_rawDescGZIP(), []int{12}
 }
 
 func (x *ForwardPaymentRequest) GetOrderId() string {
@@ -861,7 +953,7 @@ type ForwardPaymentResponse struct {
 
 func (x *ForwardPaymentResponse) Reset() {
 	*x = ForwardPaymentResponse{}
-	mi := &file_botlink_v1_botlink_proto_msgTypes[12]
+	mi := &file_botlink_v1_botlink_proto_msgTypes[13]
 	ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
 	ms.StoreMessageInfo(mi)
 }
@@ -873,7 +965,7 @@ func (x *ForwardPaymentResponse) String() string {
 func (*ForwardPaymentResponse) ProtoMessage() {}
 
 func (x *ForwardPaymentResponse) ProtoReflect() protoreflect.Message {
-	mi := &file_botlink_v1_botlink_proto_msgTypes[12]
+	mi := &file_botlink_v1_botlink_proto_msgTypes[13]
 	if x != nil {
 		ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
 		if ms.LoadMessageInfo() == nil {
@@ -886,7 +978,7 @@ func (x *ForwardPaymentResponse) ProtoReflect() protoreflect.Message {
 
 // Deprecated: Use ForwardPaymentResponse.ProtoReflect.Descriptor instead.
 func (*ForwardPaymentResponse) Descriptor() ([]byte, []int) {
-	return file_botlink_v1_botlink_proto_rawDescGZIP(), []int{12}
+	return file_botlink_v1_botlink_proto_rawDescGZIP(), []int{13}
 }
 
 func (x *ForwardPaymentResponse) GetCredited() bool {
@@ -900,11 +992,19 @@ var File_botlink_v1_botlink_proto protoreflect.FileDescriptor
 
 const file_botlink_v1_botlink_proto_rawDesc = "" +
 	"\n" +
-	"\x18botlink/v1/botlink.proto\x12\x13scrabble.botlink.v1\x1a\x1atelegram/v1/telegram.proto\"r\n" +
+	"\x18botlink/v1/botlink.proto\x12\x13scrabble.botlink.v1\x1a\x1atelegram/v1/telegram.proto\"\xa9\x01\n" +
 	"\aFromBot\x122\n" +
 	"\x05hello\x18\x01 \x01(\v2\x1a.scrabble.botlink.v1.HelloH\x00R\x05hello\x12,\n" +
-	"\x03ack\x18\x02 \x01(\v2\x18.scrabble.botlink.v1.AckH\x00R\x03ackB\x05\n" +
-	"\x03msg\"?\n" +
+	"\x03ack\x18\x02 \x01(\v2\x18.scrabble.botlink.v1.AckH\x00R\x03ack\x125\n" +
+	"\x06health\x18\x03 \x01(\v2\x1b.scrabble.botlink.v1.HealthH\x00R\x06healthB\x05\n" +
+	"\x03msg\"\x97\x01\n" +
+	"\x06Health\x12)\n" +
+	"\x10connect_failures\x18\x01 \x01(\x04R\x0fconnectFailures\x12\x1d\n" +
+	"\n" +
+	"api_errors\x18\x02 \x01(\x04R\tapiErrors\x12!\n" +
+	"\frate_limited\x18\x03 \x01(\x04R\vrateLimited\x12 \n" +
+	"\flast_ok_unix\x18\x04 \x01(\x03R\n" +
+	"lastOkUnix\"?\n" +
 	"\x05ToBot\x126\n" +
 	"\acommand\x18\x01 \x01(\v2\x1c.scrabble.botlink.v1.CommandR\acommand\"K\n" +
 	"\x05Hello\x12\x1f\n" +
@@ -976,47 +1076,49 @@ func file_botlink_v1_botlink_proto_rawDescGZIP() []byte {
 	return file_botlink_v1_botlink_proto_rawDescData
 }
 
-var file_botlink_v1_botlink_proto_msgTypes = make([]protoimpl.MessageInfo, 13)
+var file_botlink_v1_botlink_proto_msgTypes = make([]protoimpl.MessageInfo, 14)
 var file_botlink_v1_botlink_proto_goTypes = []any{
 	(*FromBot)(nil),                     // 0: scrabble.botlink.v1.FromBot
-	(*ToBot)(nil),                       // 1: scrabble.botlink.v1.ToBot
-	(*Hello)(nil),                       // 2: scrabble.botlink.v1.Hello
-	(*Command)(nil),                     // 3: scrabble.botlink.v1.Command
-	(*Ack)(nil),                         // 4: scrabble.botlink.v1.Ack
-	(*ChatGateCommand)(nil),             // 5: scrabble.botlink.v1.ChatGateCommand
-	(*ChatEligibilityRequest)(nil),      // 6: scrabble.botlink.v1.ChatEligibilityRequest
-	(*ChatEligibilityResponse)(nil),     // 7: scrabble.botlink.v1.ChatEligibilityResponse
-	(*CreateInvoiceCommand)(nil),        // 8: scrabble.botlink.v1.CreateInvoiceCommand
-	(*PreCheckoutRequest)(nil),          // 9: scrabble.botlink.v1.PreCheckoutRequest
-	(*PreCheckoutResponse)(nil),         // 10: scrabble.botlink.v1.PreCheckoutResponse
-	(*ForwardPaymentRequest)(nil),       // 11: scrabble.botlink.v1.ForwardPaymentRequest
-	(*ForwardPaymentResponse)(nil),      // 12: scrabble.botlink.v1.ForwardPaymentResponse
-	(*v1.NotifyRequest)(nil),            // 13: scrabble.telegram.v1.NotifyRequest
-	(*v1.SendToUserRequest)(nil),        // 14: scrabble.telegram.v1.SendToUserRequest
-	(*v1.SendToGameChannelRequest)(nil), // 15: scrabble.telegram.v1.SendToGameChannelRequest
+	(*Health)(nil),                      // 1: scrabble.botlink.v1.Health
+	(*ToBot)(nil),                       // 2: scrabble.botlink.v1.ToBot
+	(*Hello)(nil),                       // 3: scrabble.botlink.v1.Hello
+	(*Command)(nil),                     // 4: scrabble.botlink.v1.Command
+	(*Ack)(nil),                         // 5: scrabble.botlink.v1.Ack
+	(*ChatGateCommand)(nil),             // 6: scrabble.botlink.v1.ChatGateCommand
+	(*ChatEligibilityRequest)(nil),      // 7: scrabble.botlink.v1.ChatEligibilityRequest
+	(*ChatEligibilityResponse)(nil),     // 8: scrabble.botlink.v1.ChatEligibilityResponse
+	(*CreateInvoiceCommand)(nil),        // 9: scrabble.botlink.v1.CreateInvoiceCommand
+	(*PreCheckoutRequest)(nil),          // 10: scrabble.botlink.v1.PreCheckoutRequest
+	(*PreCheckoutResponse)(nil),         // 11: scrabble.botlink.v1.PreCheckoutResponse
+	(*ForwardPaymentRequest)(nil),       // 12: scrabble.botlink.v1.ForwardPaymentRequest
+	(*ForwardPaymentResponse)(nil),      // 13: scrabble.botlink.v1.ForwardPaymentResponse
+	(*v1.NotifyRequest)(nil),            // 14: scrabble.telegram.v1.NotifyRequest
+	(*v1.SendToUserRequest)(nil),        // 15: scrabble.telegram.v1.SendToUserRequest
+	(*v1.SendToGameChannelRequest)(nil), // 16: scrabble.telegram.v1.SendToGameChannelRequest
 }
 var file_botlink_v1_botlink_proto_depIdxs = []int32{
-	2,  // 0: scrabble.botlink.v1.FromBot.hello:type_name -> scrabble.botlink.v1.Hello
-	4,  // 1: scrabble.botlink.v1.FromBot.ack:type_name -> scrabble.botlink.v1.Ack
-	3,  // 2: scrabble.botlink.v1.ToBot.command:type_name -> scrabble.botlink.v1.Command
-	13, // 3: scrabble.botlink.v1.Command.notify:type_name -> scrabble.telegram.v1.NotifyRequest
-	14, // 4: scrabble.botlink.v1.Command.send_to_user:type_name -> scrabble.telegram.v1.SendToUserRequest
-	15, // 5: scrabble.botlink.v1.Command.send_to_channel:type_name -> scrabble.telegram.v1.SendToGameChannelRequest
-	5,  // 6: scrabble.botlink.v1.Command.chat_gate:type_name -> scrabble.botlink.v1.ChatGateCommand
-	8,  // 7: scrabble.botlink.v1.Command.create_invoice:type_name -> scrabble.botlink.v1.CreateInvoiceCommand
-	0,  // 8: scrabble.botlink.v1.BotLink.Link:input_type -> scrabble.botlink.v1.FromBot
-	6,  // 9: scrabble.botlink.v1.BotLink.ResolveChatEligibility:input_type -> scrabble.botlink.v1.ChatEligibilityRequest
-	9,  // 10: scrabble.botlink.v1.BotLink.ValidatePreCheckout:input_type -> scrabble.botlink.v1.PreCheckoutRequest
-	11, // 11: scrabble.botlink.v1.BotLink.ForwardPayment:input_type -> scrabble.botlink.v1.ForwardPaymentRequest
-	1,  // 12: scrabble.botlink.v1.BotLink.Link:output_type -> scrabble.botlink.v1.ToBot
-	7,  // 13: scrabble.botlink.v1.BotLink.ResolveChatEligibility:output_type -> scrabble.botlink.v1.ChatEligibilityResponse
-	10, // 14: scrabble.botlink.v1.BotLink.ValidatePreCheckout:output_type -> scrabble.botlink.v1.PreCheckoutResponse
-	12, // 15: scrabble.botlink.v1.BotLink.ForwardPayment:output_type -> scrabble.botlink.v1.ForwardPaymentResponse
-	12, // [12:16] is the sub-list for method output_type
-	8,  // [8:12] is the sub-list for method input_type
-	8,  // [8:8] is the sub-list for extension type_name
-	8,  // [8:8] is the sub-list for extension extendee
-	0,  // [0:8] is the sub-list for field type_name
+	3,  // 0: scrabble.botlink.v1.FromBot.hello:type_name -> scrabble.botlink.v1.Hello
+	5,  // 1: scrabble.botlink.v1.FromBot.ack:type_name -> scrabble.botlink.v1.Ack
+	1,  // 2: scrabble.botlink.v1.FromBot.health:type_name -> scrabble.botlink.v1.Health
+	4,  // 3: scrabble.botlink.v1.ToBot.command:type_name -> scrabble.botlink.v1.Command
+	14, // 4: scrabble.botlink.v1.Command.notify:type_name -> scrabble.telegram.v1.NotifyRequest
+	15, // 5: scrabble.botlink.v1.Command.send_to_user:type_name -> scrabble.telegram.v1.SendToUserRequest
+	16, // 6: scrabble.botlink.v1.Command.send_to_channel:type_name -> scrabble.telegram.v1.SendToGameChannelRequest
+	6,  // 7: scrabble.botlink.v1.Command.chat_gate:type_name -> scrabble.botlink.v1.ChatGateCommand
+	9,  // 8: scrabble.botlink.v1.Command.create_invoice:type_name -> scrabble.botlink.v1.CreateInvoiceCommand
+	0,  // 9: scrabble.botlink.v1.BotLink.Link:input_type -> scrabble.botlink.v1.FromBot
+	7,  // 10: scrabble.botlink.v1.BotLink.ResolveChatEligibility:input_type -> scrabble.botlink.v1.ChatEligibilityRequest
+	10, // 11: scrabble.botlink.v1.BotLink.ValidatePreCheckout:input_type -> scrabble.botlink.v1.PreCheckoutRequest
+	12, // 12: scrabble.botlink.v1.BotLink.ForwardPayment:input_type -> scrabble.botlink.v1.ForwardPaymentRequest
+	2,  // 13: scrabble.botlink.v1.BotLink.Link:output_type -> scrabble.botlink.v1.ToBot
+	8,  // 14: scrabble.botlink.v1.BotLink.ResolveChatEligibility:output_type -> scrabble.botlink.v1.ChatEligibilityResponse
+	11, // 15: scrabble.botlink.v1.BotLink.ValidatePreCheckout:output_type -> scrabble.botlink.v1.PreCheckoutResponse
+	13, // 16: scrabble.botlink.v1.BotLink.ForwardPayment:output_type -> scrabble.botlink.v1.ForwardPaymentResponse
+	13, // [13:17] is the sub-list for method output_type
+	9,  // [9:13] is the sub-list for method input_type
+	9,  // [9:9] is the sub-list for extension type_name
+	9,  // [9:9] is the sub-list for extension extendee
+	0,  // [0:9] is the sub-list for field type_name
 }
 
 func init() { file_botlink_v1_botlink_proto_init() }
@@ -1027,8 +1129,9 @@ func file_botlink_v1_botlink_proto_init() {
 	file_botlink_v1_botlink_proto_msgTypes[0].OneofWrappers = []any{
 		(*FromBot_Hello)(nil),
 		(*FromBot_Ack)(nil),
+		(*FromBot_Health)(nil),
 	}
-	file_botlink_v1_botlink_proto_msgTypes[3].OneofWrappers = []any{
+	file_botlink_v1_botlink_proto_msgTypes[4].OneofWrappers = []any{
 		(*Command_Notify)(nil),
 		(*Command_SendToUser)(nil),
 		(*Command_SendToChannel)(nil),
@@ -1041,7 +1144,7 @@ func file_botlink_v1_botlink_proto_init() {
 			GoPackagePath: reflect.TypeOf(x{}).PkgPath(),
 			RawDescriptor: unsafe.Slice(unsafe.StringData(file_botlink_v1_botlink_proto_rawDesc), len(file_botlink_v1_botlink_proto_rawDesc)),
 			NumEnums:      0,
-			NumMessages:   13,
+			NumMessages:   14,
 			NumExtensions: 0,
 			NumServices:   1,
 		},

pkg/proto/botlink/v1/botlink.proto

@@ -48,14 +48,30 @@ service BotLink {
 }
 
 // FromBot is a message the bot sends to the gateway: the opening Hello, then one
-// Ack per Command.
+// Ack per Command and a periodic Health report.
 message FromBot {
   oneof msg {
     Hello hello = 1;
     Ack ack = 2;
+    Health health = 3;
   }
 }
 
+// Health is a periodic report the bot pushes up the stream so the gateway can expose the remote
+// bot's Bot-API health as its own metrics — the bot exports no telemetry of its own (otelcol lives
+// on the main host, unreachable from the bot), so the bot-link is its only channel out. The three
+// counters are DELTAS since the previous Health: the gateway adds them to cumulative counters, so a
+// report lost across a reconnect only undercounts, never corrupts. last_ok_unix is the wall-clock
+// second of the bot's most recent successful Bot API response (any call, including the getUpdates
+// long-poll that returns every poll cycle even when idle) — an absolute liveness stamp the gateway
+// surfaces as a gauge, so a silently stalled bot (no errors, no traffic) is still caught.
+message Health {
+  uint64 connect_failures = 1;  // getUpdates long-poll failures (transport / 5xx) since the last report
+  uint64 api_errors = 2;        // other Bot API failures (transport / 5xx) since the last report
+  uint64 rate_limited = 3;      // Bot API 429 responses since the last report (should stay ~0)
+  int64 last_ok_unix = 4;       // unix seconds of the last successful Bot API response (0 = none yet)
+}
+
 // ToBot is a message the gateway sends to the bot. Only Command is carried today.
 message ToBot {
   Command command = 1;

platform/telegram/cmd/bot/main.go

@@ -23,6 +23,7 @@ import (
 	"scrabble/platform/telegram/internal/bot"
 	"scrabble/platform/telegram/internal/botlink"
 	"scrabble/platform/telegram/internal/config"
+	"scrabble/platform/telegram/internal/health"
 	"scrabble/platform/telegram/internal/outbox"
 	"scrabble/platform/telegram/internal/promobot"
 	"scrabble/platform/telegram/internal/support"
@@ -31,6 +32,10 @@ import (
 // telemetryShutdownTimeout bounds the OpenTelemetry flush during process exit.
 const telemetryShutdownTimeout = 5 * time.Second
 
+// healthReportInterval is how often the bot flushes its accumulated Bot API health to the gateway
+// over the bot-link. It bounds how stale a metric can be, not how often the bot talks to Telegram.
+const healthReportInterval = 30 * time.Second
+
 func main() {
 	cfg, err := config.LoadBot()
 	if err != nil {
@@ -81,6 +86,12 @@ func run(ctx context.Context, cfg config.BotConfig, logger *zap.Logger) error {
 		}
 	}
 
+	// The bot exports no telemetry of its own (otelcol is on the main host, unreachable from here),
+	// so its Bot API health is observed centrally and reported to the gateway over the bot-link,
+	// which turns the reports into metrics. Shared between the bot (which feeds it) and the bot-link
+	// client (which flushes it).
+	healthReporter := health.NewReporter()
+
 	b, err := bot.New(bot.Config{
 		Token:             cfg.Token,
 		APIBaseURL:        cfg.APIBaseURL,
@@ -92,6 +103,7 @@ func run(ctx context.Context, cfg config.BotConfig, logger *zap.Logger) error {
 		SupportChatID:     cfg.SupportChatID,
 		SupportStore:      supportStore,
 		AcceptPayments:    cfg.StarsOutboxDir != "",
+		Health:            healthReporter,
 	}, logger)
 	if err != nil {
 		return err
@@ -120,6 +132,8 @@ func run(ctx context.Context, cfg config.BotConfig, logger *zap.Logger) error {
 		OwnsUpdates:    cfg.OwnsUpdates,
 		Creds:          credentials.NewTLS(tlsCfg),
 		ReconnectDelay: cfg.BotLink.ReconnectDelay,
+		Health:         healthReporter,
+		HealthInterval: healthReportInterval,
 	}, exec, logger)
 	if err != nil {
 		return err

platform/telegram/internal/bot/bot.go

@@ -7,19 +7,26 @@ package bot
 
 import (
 	"context"
+	"net/http"
 	"net/url"
 	"strconv"
 	"strings"
+	"time"
 
 	tgbot "github.com/go-telegram/bot"
 	"github.com/go-telegram/bot/models"
 	"go.uber.org/zap"
 	"golang.org/x/time/rate"
 
+	"scrabble/platform/telegram/internal/health"
 	"scrabble/platform/telegram/internal/outbox"
 	"scrabble/platform/telegram/internal/support"
 )
 
+// botPollTimeout is the getUpdates long-poll timeout. It matches the go-telegram/bot default; we set
+// it explicitly only because wiring the health observer (WithHTTPClient) also sets the poll timeout.
+const botPollTimeout = time.Minute
+
 // Config configures the bot wrapper.
 type Config struct {
 	// Token is the Bot API token.
@@ -54,6 +61,9 @@ type Config struct {
 	// pre_checkout_query updates and handles pre_checkout / successful_payment. The runtime
 	// dependencies (the validator, forwarder and outbox) are wired with SetPaymentHandlers.
 	AcceptPayments bool
+	// Health, when set, observes every Bot API request for health metrics (reported to the gateway
+	// over the bot-link) and honours a 429's Retry-After. Nil disables the observation.
+	Health *health.Reporter
 }
 
 // EligibilityResolver answers whether the Telegram user identified by externalID
@@ -159,6 +169,12 @@ func New(cfg Config, log *zap.Logger) (*Bot, error) {
 	if cfg.APIBaseURL != "" {
 		opts = append(opts, tgbot.WithServerURL(cfg.APIBaseURL))
 	}
+	if cfg.Health != nil {
+		// Observe every Bot API request centrally for health metrics and the 429 back-off. The
+		// client timeout leaves slack over the long-poll hold (botPollTimeout - 1s server-side).
+		base := &http.Client{Timeout: botPollTimeout + 10*time.Second}
+		opts = append(opts, tgbot.WithHTTPClient(botPollTimeout, cfg.Health.Wrap(base)))
+	}
 	api, err := tgbot.New(cfg.Token, opts...)
 	if err != nil {
 		return nil, err

platform/telegram/internal/botlink/client.go

@@ -11,6 +11,7 @@ import (
 	"google.golang.org/grpc/keepalive"
 
 	botlinkv1 "scrabble/pkg/proto/botlink/v1"
+	"scrabble/platform/telegram/internal/health"
 )
 
 const (
@@ -34,6 +35,11 @@ type ClientConfig struct {
 	Creds credentials.TransportCredentials
 	// ReconnectDelay is the pause before re-dialing after the stream ends.
 	ReconnectDelay time.Duration
+	// Health, when set with a positive HealthInterval, is flushed to the gateway as a botlink Health
+	// message every HealthInterval while the stream is open. Nil disables health reporting.
+	Health *health.Reporter
+	// HealthInterval is the cadence of the Health flush; 0 disables it.
+	HealthInterval time.Duration
 }
 
 // Client maintains the long-lived bot-link to the gateway, executing the commands
@@ -140,15 +146,43 @@ func (c *Client) serve(ctx context.Context, client botlinkv1.BotLinkClient) erro
 	}
 	c.log.Info("bot-link connected", zap.String("gateway", c.cfg.GatewayAddr), zap.Bool("owns_updates", c.cfg.OwnsUpdates))
 
+	// One goroutine owns stream.Recv, feeding commands to the loop below; the loop owns every
+	// stream.Send (the Acks and the periodic Health) — a gRPC stream forbids concurrent Send.
+	rctx, cancel := context.WithCancel(ctx)
+	defer cancel()
+	cmds := make(chan *botlinkv1.Command)
+	recvErr := make(chan error, 1)
+	go func() {
 		for {
 			msg, err := stream.Recv()
 			if err != nil {
+				recvErr <- err
+				return
+			}
+			if cmd := msg.GetCommand(); cmd != nil {
+				select {
+				case cmds <- cmd:
+				case <-rctx.Done():
+					return
+				}
+			}
+		}
+	}()
+
+	var healthTick <-chan time.Time
+	if c.cfg.Health != nil && c.cfg.HealthInterval > 0 {
+		t := time.NewTicker(c.cfg.HealthInterval)
+		defer t.Stop()
+		healthTick = t.C
+	}
+
+	for {
+		select {
+		case <-ctx.Done():
+			return ctx.Err()
+		case err := <-recvErr:
 			return err
-		}
-		cmd := msg.GetCommand()
-		if cmd == nil {
-			continue
-		}
+		case cmd := <-cmds:
 			delivered, result, herr := c.exec.Handle(ctx, cmd)
 			ack := &botlinkv1.Ack{CommandId: cmd.GetCommandId(), Delivered: delivered, Result: result}
 			if herr != nil {
@@ -157,6 +191,15 @@ func (c *Client) serve(ctx context.Context, client botlinkv1.BotLinkClient) erro
 			if err := stream.Send(&botlinkv1.FromBot{Msg: &botlinkv1.FromBot_Ack{Ack: ack}}); err != nil {
 				return err
 			}
+		case <-healthTick:
+			// Snapshot without resetting; only commit (clear the deltas) after a successful send, so a
+			// failed flush loses nothing and the counts ride the next connection.
+			hb := c.cfg.Health.Snapshot()
+			if err := stream.Send(&botlinkv1.FromBot{Msg: &botlinkv1.FromBot_Health{Health: hb}}); err != nil {
+				return err
+			}
+			c.cfg.Health.Commit(hb)
+		}
 	}
 }
 

platform/telegram/internal/health/health.go

@@ -0,0 +1,145 @@
+// Package health tracks the Telegram bot's Bot API health and reports it to the gateway over the
+// bot-link. The bot exports no telemetry of its own — otelcol lives on the main host, unreachable
+// from the bot host — so the gateway turns these reports into its own metrics (see
+// gateway/internal/botlink). Health is observed CENTRALLY by wrapping the Bot API HTTP client
+// ([Reporter.Wrap]); every request is classified there with no per-call-site instrumentation.
+package health
+
+import (
+	"bytes"
+	"context"
+	"encoding/json"
+	"io"
+	"net/http"
+	"strconv"
+	"strings"
+	"sync/atomic"
+	"time"
+
+	botlinkv1 "scrabble/pkg/proto/botlink/v1"
+)
+
+// maxBackoff caps how long a single 429 makes the bot wait, so a hostile or buggy Retry-After can
+// never wedge the bot.
+const maxBackoff = 30 * time.Second
+
+// Reporter accumulates the bot's Bot API health between reports: three delta counters and the last
+// successful-response stamp. It is safe for concurrent use. The bot-link client periodically
+// [Reporter.Snapshot]s it, sends the snapshot as a botlink Health, and [Reporter.Commit]s it on a
+// successful send.
+type Reporter struct {
+	connectFailures atomic.Int64
+	apiErrors       atomic.Int64
+	rateLimited     atomic.Int64
+	lastOKUnix      atomic.Int64
+}
+
+// NewReporter returns an empty Reporter.
+func NewReporter() *Reporter { return &Reporter{} }
+
+// Snapshot reads the current delta counters and the last-ok stamp into a Health message WITHOUT
+// resetting them. The counters are cleared only by [Reporter.Commit] after a successful send, so a
+// failed send loses nothing.
+func (r *Reporter) Snapshot() *botlinkv1.Health {
+	return &botlinkv1.Health{
+		ConnectFailures: uint64(max(r.connectFailures.Load(), 0)),
+		ApiErrors:       uint64(max(r.apiErrors.Load(), 0)),
+		RateLimited:     uint64(max(r.rateLimited.Load(), 0)),
+		LastOkUnix:      r.lastOKUnix.Load(),
+	}
+}
+
+// Commit subtracts a just-sent snapshot's delta counters, so counts accrued between the snapshot and
+// the send survive into the next report. last_ok is a stamp, not a delta, so it is left as is.
+func (r *Reporter) Commit(sent *botlinkv1.Health) {
+	r.connectFailures.Add(-int64(sent.GetConnectFailures()))
+	r.apiErrors.Add(-int64(sent.GetApiErrors()))
+	r.rateLimited.Add(-int64(sent.GetRateLimited()))
+}
+
+// Wrap returns an http.Client-shaped observer over base for tgbot.WithHTTPClient: it stamps liveness
+// on every successful response, counts transport and 5xx failures (split getUpdates vs other) and
+// 429s, and honours a 429's Retry-After (bounded by maxBackoff) so the bot backs off instead of
+// hammering. A 4xx other than 429 (a user blocked the bot, a malformed request) is a normal
+// per-request outcome, not a health signal, and is not counted.
+func (r *Reporter) Wrap(base Doer) Doer { return &observer{base: base, r: r} }
+
+// Doer is the minimal HTTP surface the Bot API client needs; both *http.Client and [observer]
+// satisfy it, and it matches the go-telegram/bot HttpClient interface.
+type Doer interface {
+	Do(*http.Request) (*http.Response, error)
+}
+
+// observer is the Bot API HTTP client wrapper (see [Reporter.Wrap]).
+type observer struct {
+	base Doer
+	r    *Reporter
+}
+
+// Do performs the request and records its health outcome.
+func (o *observer) Do(req *http.Request) (*http.Response, error) {
+	resp, err := o.base.Do(req)
+	poll := strings.HasSuffix(req.URL.Path, "/getUpdates")
+	if err != nil {
+		// A cancelled context is a shutdown, not a Bot API failure.
+		if req.Context().Err() == nil {
+			o.fail(poll)
+		}
+		return resp, err
+	}
+	switch {
+	case resp.StatusCode == http.StatusTooManyRequests:
+		o.r.rateLimited.Add(1)
+		o.backoff(req.Context(), resp)
+	case resp.StatusCode >= 200 && resp.StatusCode < 300:
+		o.r.lastOKUnix.Store(time.Now().Unix())
+	case resp.StatusCode >= 500:
+		o.fail(poll)
+	}
+	return resp, err
+}
+
+// fail records a connect failure on the getUpdates long-poll or an api error otherwise.
+func (o *observer) fail(poll bool) {
+	if poll {
+		o.r.connectFailures.Add(1)
+	} else {
+		o.r.apiErrors.Add(1)
+	}
+}
+
+// backoff waits out a 429's Retry-After (bounded, cancellable). It buffers the response body first so
+// the wait holds no connection open and the Bot API library can still parse the 429 afterwards.
+func (o *observer) backoff(ctx context.Context, resp *http.Response) {
+	body, _ := io.ReadAll(resp.Body)
+	resp.Body.Close()
+	resp.Body = io.NopCloser(bytes.NewReader(body))
+	d := retryAfter(resp.Header.Get("Retry-After"), body)
+	if d <= 0 {
+		return
+	}
+	if d > maxBackoff {
+		d = maxBackoff
+	}
+	select {
+	case <-ctx.Done():
+	case <-time.After(d):
+	}
+}
+
+// retryAfter resolves the 429 back-off from the Retry-After header, falling back to the Bot API
+// body's parameters.retry_after (both are seconds). It returns 0 when neither gives a positive value.
+func retryAfter(header string, body []byte) time.Duration {
+	if secs, err := strconv.Atoi(strings.TrimSpace(header)); err == nil && secs > 0 {
+		return time.Duration(secs) * time.Second
+	}
+	var payload struct {
+		Parameters struct {
+			RetryAfter int `json:"retry_after"`
+		} `json:"parameters"`
+	}
+	if err := json.Unmarshal(body, &payload); err == nil && payload.Parameters.RetryAfter > 0 {
+		return time.Duration(payload.Parameters.RetryAfter) * time.Second
+	}
+	return 0
+}

platform/telegram/internal/health/health_test.go

@@ -0,0 +1,121 @@
+package health
+
+import (
+	"context"
+	"errors"
+	"io"
+	"net/http"
+	"net/http/httptest"
+	"strings"
+	"testing"
+	"time"
+)
+
+// fakeDoer returns a canned response and error for every request.
+type fakeDoer struct {
+	resp *http.Response
+	err  error
+}
+
+func (f *fakeDoer) Do(*http.Request) (*http.Response, error) { return f.resp, f.err }
+
+// mkResp builds a response with a status, optional Retry-After header and body.
+func mkResp(status int, retryAfter, body string) *http.Response {
+	h := http.Header{}
+	if retryAfter != "" {
+		h.Set("Retry-After", retryAfter)
+	}
+	return &http.Response{StatusCode: status, Header: h, Body: io.NopCloser(strings.NewReader(body))}
+}
+
+// run sends one request for the Bot API method through a fresh observer and returns the reporter's
+// resulting snapshot. The 429 cases carry no Retry-After, so the back-off returns before any wait.
+func run(method string, resp *http.Response, err error) *Reporter {
+	r := NewReporter()
+	req := httptest.NewRequest(http.MethodPost, "https://api.telegram.org/bot123/"+method, nil)
+	_, _ = r.Wrap(&fakeDoer{resp: resp, err: err}).Do(req)
+	return r
+}
+
+func TestObserverClassifies(t *testing.T) {
+	cases := []struct {
+		name          string
+		method        string
+		resp          *http.Response
+		err           error
+		wantConnect   uint64
+		wantAPI       uint64
+		wantRate      uint64
+		wantLastOKSet bool
+	}{
+		{"poll ok stamps liveness", "getUpdates", mkResp(200, "", `{"ok":true}`), nil, 0, 0, 0, true},
+		{"send ok stamps liveness", "sendMessage", mkResp(200, "", `{"ok":true}`), nil, 0, 0, 0, true},
+		{"poll 5xx is a connect failure", "getUpdates", mkResp(502, "", ""), nil, 1, 0, 0, false},
+		{"send 5xx is an api error", "sendMessage", mkResp(500, "", ""), nil, 0, 1, 0, false},
+		{"429 is rate limited, not an error", "sendMessage", mkResp(429, "", `{"ok":false}`), nil, 0, 0, 1, false},
+		{"4xx other than 429 is not counted", "sendMessage", mkResp(403, "", ""), nil, 0, 0, 0, false},
+		{"poll transport error is a connect failure", "getUpdates", nil, errors.New("dial"), 1, 0, 0, false},
+		{"send transport error is an api error", "sendMessage", nil, errors.New("dial"), 0, 1, 0, false},
+	}
+	for _, tc := range cases {
+		t.Run(tc.name, func(t *testing.T) {
+			s := run(tc.method, tc.resp, tc.err).Snapshot()
+			if s.GetConnectFailures() != tc.wantConnect || s.GetApiErrors() != tc.wantAPI || s.GetRateLimited() != tc.wantRate {
+				t.Errorf("counters = connect %d api %d rate %d, want %d/%d/%d",
+					s.GetConnectFailures(), s.GetApiErrors(), s.GetRateLimited(), tc.wantConnect, tc.wantAPI, tc.wantRate)
+			}
+			if (s.GetLastOkUnix() > 0) != tc.wantLastOKSet {
+				t.Errorf("last_ok set = %v, want %v", s.GetLastOkUnix() > 0, tc.wantLastOKSet)
+			}
+		})
+	}
+}
+
+// TestObserverIgnoresShutdownTransportError checks a transport error under a cancelled context (a
+// shutdown) is not counted as a Bot API failure.
+func TestObserverIgnoresShutdownTransportError(t *testing.T) {
+	r := NewReporter()
+	req := httptest.NewRequest(http.MethodPost, "https://api.telegram.org/bot123/getUpdates", nil)
+	ctx, cancel := context.WithCancel(req.Context())
+	cancel()
+	_, _ = r.Wrap(&fakeDoer{err: errors.New("dial")}).Do(req.WithContext(ctx))
+	if s := r.Snapshot(); s.GetConnectFailures() != 0 {
+		t.Errorf("shutdown transport error must not count, got connect=%d", s.GetConnectFailures())
+	}
+}
+
+func TestRetryAfter(t *testing.T) {
+	cases := []struct {
+		header string
+		body   string
+		want   time.Duration
+	}{
+		{"5", "", 5 * time.Second},
+		{"", `{"parameters":{"retry_after":7}}`, 7 * time.Second},
+		{"3", `{"parameters":{"retry_after":9}}`, 3 * time.Second}, // header wins
+		{"", `{"ok":false}`, 0},
+		{"0", "", 0},
+		{"", "not json", 0},
+	}
+	for _, tc := range cases {
+		if got := retryAfter(tc.header, []byte(tc.body)); got != tc.want {
+			t.Errorf("retryAfter(%q, %q) = %v, want %v", tc.header, tc.body, got, tc.want)
+		}
+	}
+}
+
+// TestSnapshotCommit checks Commit clears only the sent deltas, so counts accrued between the
+// snapshot and the send survive into the next report.
+func TestSnapshotCommit(t *testing.T) {
+	r := NewReporter()
+	r.apiErrors.Add(3)
+	snap := r.Snapshot()
+	if snap.GetApiErrors() != 3 {
+		t.Fatalf("snapshot api_errors = %d, want 3", snap.GetApiErrors())
+	}
+	r.apiErrors.Add(2) // accrues after the snapshot
+	r.Commit(snap)
+	if got := r.Snapshot().GetApiErrors(); got != 2 {
+		t.Errorf("after commit api_errors = %d, want 2 (the post-snapshot accrual)", got)
+	}
+}