diff --git a/tasks/ha-availability-dashboard/appdaemon/apps.yaml b/tasks/ha-availability-dashboard/appdaemon/apps.yaml
new file mode 100644
index 0000000..b49911c
--- /dev/null
+++ b/tasks/ha-availability-dashboard/appdaemon/apps.yaml
@@ -0,0 +1,11 @@
+# AppDaemon apps configuration
+# Add the availability app alongside existing apps (e.g., hello_world)
+
+hello_world:
+  module: hello
+  class: HelloWorld
+
+availability:
+  module: availability
+  class: Availability
+  log_level: info
diff --git a/tasks/ha-availability-dashboard/appdaemon/availability.py b/tasks/ha-availability-dashboard/appdaemon/availability.py
new file mode 100644
index 0000000..db8c3c3
--- /dev/null
+++ b/tasks/ha-availability-dashboard/appdaemon/availability.py
@@ -0,0 +1,322 @@
+"""AppDaemon app: Device availability calculator for Home Assistant.
+
+Calculates uptime metrics for tracked devices and creates sensor.avail_* entities.
+"""
+
+import appdaemon.plugins.hass.hassapi as hass
+import requests
+from datetime import datetime, timezone, timedelta
+from availability_utils import (
+    is_excluded, sanitize_entity_id, sanitize_area_name,
+    get_color, calc_trend, format_downtime,
+    parse_datetime, period_timedelta, chunk_list,
+    compute_availability, compute_sparkline,
+    TRACKED_DOMAINS,
+)
+
+BATCH_SIZE = 20
+BATCH_DELAY = 1  # seconds between batch requests
+HA_URL = "http://supervisor/core/api"
+
+
+class Availability(hass.Hass):
+    def initialize(self):
+        self._token = None
+        self._entities_cache = []
+        self._areas_cache = {}
+        self._current_period = "7d"
+        self._results_cache = {"24h": {}, "7d": {}, "30d": {}}
+
+        # Cold-start: full recalc after 30s
+        self.run_in(self._cold_start, 30)
+
+        # Periodic schedules
+        self.run_every(self._calc_24h, "now+60", 5 * 60)
+        self.run_every(self._calc_7d, "now+90", 15 * 60)
+        self.run_every(self._calc_30d, "now+120", 2 * 3600)
+
+        # Listen for period selector change
+        self.listen_state(self._period_changed, "input_select.avail_period")
+
+        self.log("Availability app initialized")
+
+    def _get_token(self):
+        if self._token is None:
+            import os
+            self._token = os.environ.get("SUPERVISOR_TOKEN", "")
+            if not self._token:
+                self._token = self.args.get("token", "")
+        return self._token
+
+    def _api_headers(self):
+        return {
+            "Authorization": f"Bearer {self._get_token()}",
+            "Content-Type": "application/json",
+        }
+
+    def _api_get(self, path):
+        url = f"{HA_URL}{path}"
+        try:
+            resp = requests.get(url, headers=self._api_headers(), timeout=30)
+            resp.raise_for_status()
+            return resp.json()
+        except Exception as e:
+            self.log(f"API error {path}: {e}", level="WARNING")
+            return None
+
+    # ── Entity Discovery ──
+
+    def _fetch_entities(self):
+        """Get all entities from HA and filter by domain + exclusions."""
+        data = self._api_get("/states")
+        if not data:
+            self.log("Failed to fetch entities", level="ERROR")
+            return []
+
+        entities = []
+        for e in data:
+            eid = e.get("entity_id", "")
+            domain = eid.split(".")[0] if "." in eid else ""
+            if domain not in TRACKED_DOMAINS:
+                continue
+            if is_excluded(eid):
+                continue
+            entities.append({
+                "entity_id": eid,
+                "friendly_name": e.get("attributes", {}).get("friendly_name", eid),
+                "domain": domain,
+            })
+        self._entities_cache = entities
+        self.log(f"Found {len(entities)} tracked entities")
+        return entities
+
+    def _fetch_areas(self):
+        """Fetch area registry and build entity_id -> area mapping."""
+        data = self._api_get("/config/area_registry")
+        if not data:
+            self._areas_cache = {}
+            return {}
+
+        # Build device_id -> area map
+        device_areas = {}
+        for area in data:
+            area_id = area.get("area_id")
+            area_name = area.get("name", "Без комнаты")
+            device_areas[area_id] = area_name
+
+        # Now get entity registry to map entity -> device -> area
+        reg = self._api_get("/config/entity_registry")
+        if not reg:
+            self._areas_cache = {}
+            return {}
+
+        eid_to_area = {}
+        for entry in reg:
+            eid = entry.get("entity_id", "")
+            device_id = entry.get("device_id")
+            area_id = entry.get("area_id")
+            # Use entity-level area if set, otherwise device-level
+            if area_id and area_id in device_areas:
+                eid_to_area[eid] = device_areas[area_id]
+            elif device_id:
+                # Try to find device's area via device registry
+                dev_data = self._api_get(f"/config/device_registry")
+                # This is expensive; fallback to simple mapping
+                pass
+            else:
+                eid_to_area[eid] = "Без комнаты"
+
+        # Fill from device registry if available
+        dev_reg = self._api_get("/config/device_registry")
+        if dev_reg:
+            dev_area_map = {}
+            for dev in dev_reg:
+                dev_id = dev.get("id")
+                a_id = dev.get("area_id")
+                if dev_id and a_id and a_id in device_areas:
+                    dev_area_map[dev_id] = device_areas[a_id]
+
+            for entry in reg:
+                eid = entry.get("entity_id", "")
+                device_id = entry.get("device_id")
+                area_id = entry.get("area_id")
+                if area_id and area_id in device_areas:
+                    eid_to_area[eid] = device_areas[area_id]
+                elif device_id and device_id in dev_area_map:
+                    eid_to_area[eid] = dev_area_map[device_id]
+                else:
+                    eid_to_area[eid] = eid_to_area.get(eid, "Без комнаты")
+
+        self._areas_cache = eid_to_area
+        return eid_to_area
+
+    # ── History Fetch ──
+
+    def _fetch_history(self, entity_ids: list, period_start: datetime, period_end: datetime):
+        """Fetch history from HA API in batches."""
+        start_str = period_start.isoformat()
+        all_history = {}
+
+        chunks = chunk_list(entity_ids, BATCH_SIZE)
+        total_chunks = len(chunks)
+
+        for i, chunk in enumerate(chunks):
+            ids_str = ",".join(chunk)
+            path = f"/history/period/{start_str}?filter_entity_id={ids_str}&minimal_response&no_attributes"
+            data = self._api_get(path)
+
+            if data and isinstance(data, list):
+                for entity_history in data:
+                    if entity_history and isinstance(entity_history, list) and len(entity_history) > 0:
+                        eid = entity_history[0].get("entity_id", chunk[0] if chunk else "")
+                        all_history[eid] = entity_history
+
+            # Update progress
+            progress = f"{i + 1}/{total_chunks}"
+            self._set_progress(progress)
+
+            if i < total_chunks - 1:
+                self.sleep(BATCH_DELAY)
+
+        self._set_progress("idle")
+        return all_history
+
+    # ── Calculation ──
+
+    def _calc_period(self, period: str):
+        """Calculate availability for all devices in given period."""
+        self.log(f"Calculating availability for period {period}")
+
+        now = datetime.now(timezone.utc)
+        td = period_timedelta(period)
+        period_start = now - td
+        period_end = now
+
+        # Previous period for trend
+        prev_start = period_start - td
+        prev_end = period_start
+
+        entities = self._fetch_entities()
+        if not entities:
+            return
+
+        areas = self._fetch_areas()
+        entity_ids = [e["entity_id"] for e in entities]
+
+        # Fetch history for current period
+        history = self._fetch_history(entity_ids, period_start, period_end)
+
+        # Fetch history for previous period (for trend)
+        prev_history = self._fetch_history(entity_ids, prev_start, prev_end)
+
+        results = {}
+        area_results = {}
+
+        for ent in entities:
+            eid = ent["entity_id"]
+            fname = ent["friendly_name"]
+            domain = ent["domain"]
+            area = areas.get(eid, "Без комнаты")
+
+            entries = history.get(eid, [])
+            metrics = compute_availability(entries, period_start, period_end)
+
+            # Sparkline
+            spark_days = 7 if period in ("7d", "30d") else 1
+            sparkline = compute_sparkline(entries, period_end, days=spark_days)
+
+            # Trend
+            prev_entries = prev_history.get(eid, [])
+            prev_metrics = compute_availability(prev_entries, prev_start, prev_end)
+            trend = calc_trend(metrics["availability_pct"], prev_metrics["availability_pct"])
+
+            color = get_color(metrics["availability_pct"])
+
+            result = {
+                "entity_id": eid,
+                "friendly_name": fname,
+                "domain": domain,
+                "area": area,
+                "period": period,
+                "availability_pct": metrics["availability_pct"],
+                "down_count": metrics["down_count"],
+                "max_downtime_minutes": metrics["max_downtime_minutes"],
+                "sparkline": sparkline,
+                "trend": trend,
+                "last_downtime": metrics["last_downtime"],
+                "color": color,
+                "last_updated": now.isoformat(),
+            }
+            results[eid] = result
+
+            # Area aggregation
+            if area not in area_results:
+                area_results[area] = {"pcts": [], "devices": 0, "problems": 0}
+            area_results[area]["pcts"].append(metrics["availability_pct"])
+            area_results[area]["devices"] += 1
+            if metrics["availability_pct"] < 95:
+                area_results[area]["problems"] += 1
+
+        self._results_cache[period] = results
+
+        # Write device sensors
+        for eid, res in results.items():
+            safe = sanitize_entity_id(eid)
+            sensor_id = f"sensor.avail_{safe}"
+            attrs = {k: v for k, v in res.items() if k != "state"}
+            self.set_state(sensor_id, state=str(res["availability_pct"]), attributes=attrs)
+
+        # Write area sensors
+        for area_name, adata in area_results.items():
+            avg = round(sum(adata["pcts"]) / len(adata["pcts"]), 1) if adata["pcts"] else 100.0
+            safe_area = sanitize_area_name(area_name)
+            sensor_id = f"sensor.avail_area_{safe_area}"
+            attrs = {
+                "area": area_name,
+                "period": period,
+                "availability_pct": avg,
+                "device_count": adata["devices"],
+                "problem_count": adata["problems"],
+                "color": get_color(avg),
+                "last_updated": now.isoformat(),
+            }
+            self.set_state(sensor_id, state=str(avg), attributes=attrs)
+
+        self.log(f"Calculated {len(results)} device sensors, {len(area_results)} area sensors for {period}")
+
+    # ── Callbacks ──
+
+    def _cold_start(self, kwargs):
+        """Initial calculation on startup."""
+        self.log("Cold start: calculating current period")
+        try:
+            period = self.get_state("input_select.avail_period") or "7d"
+        except Exception:
+            period = "7d"
+        self._current_period = period
+        self._calc_period(period)
+
+    def _calc_24h(self, kwargs):
+        self._calc_period("24h")
+
+    def _calc_7d(self, kwargs):
+        self._calc_period("7d")
+
+    def _calc_30d(self, kwargs):
+        self._calc_period("30d")
+
+    def _period_changed(self, entity, attribute, old, new, kwargs):
+        """Handle period selector change."""
+        if new and new != old:
+            self._current_period = new
+            self.log(f"Period changed to {new}, recalculating")
+            self._calc_period(new)
+
+    # ── Progress ──
+
+    def _set_progress(self, value: str):
+        """Update calculation progress sensor."""
+        self.set_state("sensor.avail_calc_progress", state=value, attributes={
+            "friendly_name": "Расчёт доступности",
+            "last_updated": datetime.now(timezone.utc).isoformat(),
+        })
diff --git a/tasks/ha-availability-dashboard/appdaemon/availability_utils.py b/tasks/ha-availability-dashboard/appdaemon/availability_utils.py
new file mode 100644
index 0000000..3ad9a31
--- /dev/null
+++ b/tasks/ha-availability-dashboard/appdaemon/availability_utils.py
@@ -0,0 +1,266 @@
+"""Utility functions for availability calculator."""
+
+import re
+from datetime import datetime, timezone, timedelta
+
+# Domains to track (phase 1)
+TRACKED_DOMAINS = {"light", "switch"}
+
+# Entity suffixes to exclude (Zigbee2MQTT helpers, etc.)
+EXCLUDE_SUFFIXES = (
+    "_battery_low", "_battery", "_linkquality", "_update",
+    "_update_state", "_identify",
+)
+
+# Additional exclusions for switch domain (relay settings, not real devices)
+SWITCH_EXCLUDE_SUFFIXES = (
+    "_delayed_power_on_state", "_detach_relay_mode",
+    "_network_indicator", "_turbo_mode", "_do_not_disturb",
+)
+
+SWITCH_EXCLUDE_EXACT = {"switch.zigbee2mqtt_bridge_permit_join"}
+
+# Auto-generated Zigbee select/number patterns
+ZIGBEE_AUTO_PATTERN = re.compile(
+    r"(select|number)\.(zigbee2mqtt_|.*_(radar_sensitivity|sensitivity|mode|level|delay|threshold|interval|timeout))",
+    re.IGNORECASE,
+)
+
+# Update/button domain patterns
+UPDATE_BUTTON_EXCLUDE = re.compile(r"^(update|button)\.", re.IGNORECASE)
+
+
+def is_excluded(entity_id: str) -> bool:
+    """Check if entity should be excluded from tracking."""
+    domain = entity_id.split(".")[0]
+    name = entity_id.split(".")[1] if "." in entity_id else ""
+
+    # Exact exclusions
+    if entity_id in SWITCH_EXCLUDE_EXACT:
+        return True
+
+    # Suffix exclusions
+    for suffix in EXCLUDE_SUFFIXES:
+        if name.endswith(suffix):
+            return True
+
+    # Switch-specific suffix exclusions
+    if domain == "switch":
+        for suffix in SWITCH_EXCLUDE_SUFFIXES:
+            if name.endswith(suffix):
+                return True
+
+    return False
+
+
+def sanitize_entity_id(entity_id: str) -> str:
+    """Convert entity_id to safe sensor name: light.bra_v_spalne → light_bra_v_spalne."""
+    return entity_id.replace(".", "_")
+
+
+def sanitize_area_name(name: str) -> str:
+    """Convert area name to safe sensor suffix."""
+    s = name.lower().strip()
+    s = re.sub(r"[^a-zа-яё0-9_]", "_", s)
+    s = re.sub(r"_+", "_", s)
+    return s.strip("_")
+
+
+def domain_icon(domain: str) -> str:
+    """Return icon for domain."""
+    return {
+        "light": "💡",
+        "switch": "🔌",
+        "binary_sensor": "👁️",
+        "sensor": "🌡️",
+        "climate": "❄️",
+        "cover": "🚪",
+        "lock": "🔒",
+        "media_player": "🔊",
+        "device_tracker": "📍",
+        "vacuum": "🤖",
+        "fan": "🌀",
+        "humidifier": "💨",
+        "water_heater": "🚿",
+        "siren": "🚨",
+        "button": "🔘",
+    }.get(domain, "📡")
+
+
+def get_color(pct: float) -> str:
+    """Return color bucket for availability percentage."""
+    if pct >= 99:
+        return "green"
+    elif pct >= 95:
+        return "yellow"
+    elif pct >= 90:
+        return "orange"
+    return "red"
+
+
+def calc_trend(current_pct: float, previous_pct: float, threshold: float = 0.5) -> str:
+    """Compare current vs previous period. Returns up/down/stable."""
+    diff = current_pct - previous_pct
+    if diff > threshold:
+        return "up"
+    elif diff < -threshold:
+        return "down"
+    return "stable"
+
+
+def format_downtime(minutes: int) -> str:
+    """Format downtime minutes to human-readable string."""
+    if minutes < 60:
+        return f"{minutes} мин"
+    h, m = divmod(minutes, 60)
+    if h < 24:
+        return f"{h}ч {m}мин"
+    d, h = divmod(h, 24)
+    return f"{d}д {h}ч {m}мин"
+
+
+def parse_datetime(dt_str: str) -> datetime:
+    """Parse HA datetime string to timezone-aware datetime."""
+    # HA format: 2026-04-14T15:32:00+03:00 or 2026-04-14T15:32:00.123456+03:00
+    dt_str = dt_str.rstrip("Z")
+    if "+" in dt_str[10:] or dt_str.endswith("Z"):
+        # Has timezone
+        if "+" in dt_str:
+            # Python fromisoformat handles this in 3.7+
+            return datetime.fromisoformat(dt_str)
+        return datetime.fromisoformat(dt_str.replace("Z", "+00:00"))
+    # No timezone — assume UTC
+    dt = datetime.fromisoformat(dt_str)
+    return dt.replace(tzinfo=timezone.utc)
+
+
+def period_timedelta(period: str) -> timedelta:
+    """Convert period string to timedelta."""
+    return {"24h": timedelta(hours=24), "7d": timedelta(days=7), "30d": timedelta(days=30)}[period]
+
+
+def chunk_list(lst: list, size: int) -> list:
+    """Split list into chunks of given size."""
+    return [lst[i:i + size] for i in range(0, len(lst), size)]
+
+
+def compute_availability(history_entries: list, period_start: datetime, period_end: datetime) -> dict:
+    """
+    Compute availability metrics from HA history entries.
+
+    Each entry is a dict with 'state' and 'last_changed'.
+    Returns dict with: availability_pct, down_count, max_downtime_minutes, last_downtime
+    """
+    unavailable_seconds = 0
+    down_count = 0
+    max_downtime = 0
+    current_downtime_start = None
+    last_downtime = None
+    total_seconds = (period_end - period_start).total_seconds()
+
+    if not history_entries or total_seconds <= 0:
+        return {
+            "availability_pct": 100.0,
+            "down_count": 0,
+            "max_downtime_minutes": 0,
+            "last_downtime": None,
+        }
+
+    # Sort by last_changed
+    entries = sorted(history_entries, key=lambda e: parse_datetime(e["last_changed"]))
+
+    for i, entry in enumerate(entries):
+        state = entry["state"]
+        changed = parse_datetime(entry["last_changed"])
+
+        # Clamp to period
+        if changed < period_start:
+            changed = period_start
+
+        # Determine next change time
+        if i + 1 < len(entries):
+            next_changed = parse_datetime(entries[i + 1]["last_changed"])
+        else:
+            next_changed = period_end
+
+        if next_changed > period_end:
+            next_changed = period_end
+
+        if state in ("unavailable", "unknown"):
+            if current_downtime_start is None:
+                # Transition to unavailable
+                current_downtime_start = changed
+                down_count += 1
+
+            duration = (next_changed - changed).total_seconds()
+            if duration > 0:
+                unavailable_seconds += duration
+                last_downtime = changed
+
+            # Check if this is the end of a downtime period
+            if i + 1 < len(entries) and entries[i + 1]["state"] not in ("unavailable", "unknown"):
+                if current_downtime_start is not None:
+                    downtime = (next_changed - current_downtime_start).total_seconds()
+                    if downtime > max_downtime:
+                        max_downtime = downtime
+                    current_downtime_start = None
+            # If last entry and still unavailable
+            elif i == len(entries) - 1:
+                if current_downtime_start is not None:
+                    downtime = (next_changed - current_downtime_start).total_seconds()
+                    if downtime > max_downtime:
+                        max_downtime = downtime
+                    current_downtime_start = None
+        else:
+            # Available state — if there was a downtime running, end it
+            if current_downtime_start is not None:
+                downtime = (changed - current_downtime_start).total_seconds()
+                if downtime > max_downtime:
+                    max_downtime = downtime
+                current_downtime_start = None
+
+    # Handle case where device was unavailable at period_start
+    # (first entry state determines status from period_start to first change)
+    if entries and entries[0]["state"] in ("unavailable", "unknown"):
+        first_change = parse_datetime(entries[0]["last_changed"])
+        if first_change > period_start:
+            pre_duration = (first_change - period_start).total_seconds()
+            unavailable_seconds += pre_duration
+            # This counts as one transition if not already counted
+            if current_downtime_start is None:
+                down_count += 1
+
+    # Handle trailing unavailable (last entry is unavailable)
+    if entries and entries[-1]["state"] in ("unavailable", "unknown"):
+        last_change = parse_datetime(entries[-1]["last_changed"])
+        trailing = (period_end - last_change).total_seconds()
+        if trailing > 0 and current_downtime_start is None:
+            # Already counted in the loop
+            pass
+
+    pct = round((1 - unavailable_seconds / total_seconds) * 100, 1) if total_seconds > 0 else 100.0
+    pct = max(0.0, min(100.0, pct))
+
+    return {
+        "availability_pct": pct,
+        "down_count": down_count,
+        "max_downtime_minutes": round(max_downtime / 60),
+        "last_downtime": last_downtime.isoformat() if last_downtime else None,
+    }
+
+
+def compute_sparkline(history_entries: list, period_end: datetime, days: int = 7) -> list:
+    """Compute daily availability percentages for sparkline (last N days)."""
+    points = []
+    for i in range(days - 1, -1, -1):
+        day_start = period_end - timedelta(days=i + 1)
+        day_end = period_end - timedelta(days=i)
+        # Filter entries relevant to this day
+        day_entries = []
+        for e in history_entries:
+            changed = parse_datetime(e["last_changed"])
+            if changed < day_end and changed >= day_start - timedelta(days=1):
+                day_entries.append(e)
+        result = compute_availability(day_entries, day_start, day_end)
+        points.append(result["availability_pct"])
+    return points