fix(connectivity): drive reachability from real repository traffic

The offline/online switch was janky because two independent systems decided
"online" and never communicated:

- ConnectivityMonitor owned is_server_reachable (drove the UI banner) but
  learned reachability only from a standalone /System/Info/Public ping loop
  and from auth/login calls.
- HybridRepository served all real data by racing cache-vs-server but never
  read or wrote reachability.

So the banner reflected a side-channel poller, not the system the user actually
experienced: a successful ping could read "online" while authenticated data
calls 401'd or timed out, and three different timeout regimes (5s ping / 30s
data / 100ms cache race) flapped against each other.

Unify into a single source of truth:

- Extract a cheap, cloneable ConnectivityReporter that owns all reachability
  transitions and event emission.
- OnlineRepository reports the outcome of every server request to the reporter,
  classified via RepoError: Ok/Authentication/NotFound/Server => reachable
  (the server answered), Network => offline candidate, Database/Offline =>
  ignored (not a server signal).
- Time-window debounce (OFFLINE_CONFIRM_WINDOW = 5s): flip offline only after
  sustained network failure; recover instantly on the first success.
- Demote the ping loop to an offline-only recovery probe (no online polling;
  real traffic is the signal when online).
- Frontend: navigator.onLine is now advisory (triggers a recheck instead of
  forcing offline); removed the dead markReachable/markUnreachable store methods.

Docs updated (README, 07-connectivity, 03-data-flow, 02-svelte-frontend) to
describe the new model and fix pre-existing drift (HTTP client is 30s timeout +
5s ping, not the documented 10s/base_url).

Tests: 12 connectivity tests (debounce, instant recovery, RepoError
classification through report_outcome). Full suite: 398 Rust + 384 frontend
passing, svelte-check clean.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

docs/architecture/02-svelte-frontend.md

@@ -176,7 +176,10 @@ classDiagram
         -server_url: String
         -user_id: String
         -access_token: String
+        -connectivity: Option~Arc~ConnectivityMonitor~~
         +new()
+        +with_connectivity()
+        -report_outcome()
     }
 
     class OfflineRepository {
@@ -192,10 +195,9 @@ classDiagram
     class HybridRepository {
         -online: Arc~OnlineRepository~
         -offline: Arc~OfflineRepository~
-        -connectivity: Arc~ConnectivityMonitor~
         +new()
-        -parallel_query()
-        -has_meaningful_content()
+        -parallel_race()
+        -cache_with_timeout()
     }
 
     MediaRepository <|.. OnlineRepository
@@ -214,6 +216,7 @@ classDiagram
    - Returns cache result if it has meaningful content
    - Falls back to server result otherwise
    - Background cache updates planned
+   - **Connectivity feedback**: `OnlineRepository` reports the outcome of every server request to the `ConnectivityMonitor` (classified via `RepoError`). This is the source of truth for the offline/online banner — see [07-connectivity.md](07-connectivity.md). The frontend `connectivity` store is a pure reflection of the resulting events; `navigator.onLine` is only an advisory hint that triggers an immediate recheck.
 
 2. **Handle-Based Resource Management** (`repository.rs` commands):
    ```rust

docs/architecture/03-data-flow.md

@@ -10,6 +10,7 @@ sequenceDiagram
     participant Hybrid as HybridRepository
     participant Cache as OfflineRepository (SQLite)
     participant Server as OnlineRepository (HTTP)
+    participant Conn as ConnectivityMonitor
 
     UI->>Client: getItems(parentId)
     Client->>Rust: invoke("repository_get_items", {handle, parentId})
@@ -20,6 +21,13 @@ sequenceDiagram
         Hybrid->>Server: get_items() (no timeout)
     end
 
+    Note over Server,Conn: Every server request reports its outcome
+    alt Server succeeds (or answers with 4xx/5xx)
+        Server->>Conn: mark_reachable() (server is up)
+    else Network failure / timeout
+        Server->>Conn: mark_unreachable() (debounced)
+    end
+
     alt Cache returns with content
         Cache-->>Hybrid: Result with items
         Hybrid-->>Rust: Return cache result
@@ -39,6 +47,7 @@ sequenceDiagram
 - Cache wins if it has meaningful content
 - Automatic fallback to server if cache is empty/stale
 - Background cache updates (planned)
+- **Connectivity side-effect**: each server request feeds the `ConnectivityMonitor`, which is the source of truth for the offline/online banner (see [07-connectivity.md](07-connectivity.md)). A server-answered error (401/404/5xx) still counts as *reachable* — only network failures, sustained past a debounce window, flip the app to offline.
 
 ## Playback Initiation Flow
 

docs/architecture/07-connectivity.md

@@ -13,12 +13,13 @@ pub struct HttpClient {
 }
 
 pub struct HttpConfig {
-    pub base_url: String,
-    pub timeout: Duration,        // Default: 10s
+    pub timeout: Duration,        // Default: 30s (large library queries can be slow)
     pub max_retries: u32,         // Default: 3
 }
 ```
 
+> Note: ordinary requests use the 30s timeout above. The connectivity recovery probe (`ping`) uses a shorter, dedicated 5s timeout so an unreachable server is detected quickly while offline.
+
 **Retry Strategy:**
 - Retry delays: 1s, 2s, 4s (exponential backoff)
 - Retries on: Network errors, 5xx server errors
@@ -38,39 +39,71 @@ pub enum ErrorKind {
 
 **Location**: `src-tauri/src/connectivity/mod.rs`
 
-The connectivity monitor tracks server reachability with adaptive polling:
+The connectivity monitor is the **single source of truth** for server reachability. Its primary signal is the outcome of *real repository traffic* — every server request the user actually makes. A standalone `/System/Info/Public` probe is kept only as an offline recovery detector.
+
+### Source of truth: repository traffic
+
+`OnlineRepository` reports the result of each server request to the monitor, classified via `RepoError`:
+
+| Repository outcome | Meaning | Effect on reachability |
+|--------------------|---------|------------------------|
+| `Ok(_)` | Server answered successfully | Mark **reachable** (instant recovery) |
+| `Err(Authentication)` | Server answered with 401/403 | Mark **reachable** (server is up; request was rejected) |
+| `Err(NotFound)` | Server answered with 404 | Mark **reachable** (server is up) |
+| `Err(Server)` | Server answered with 5xx / bad body | Mark **reachable** (server is up) |
+| `Err(Network)` | Connection failure / timeout / DNS | **Candidate for offline** (see debounce) |
+| `Err(Database)` | Local cache error only | No effect (not a server signal) |
+
+This classification fixes the previous bug where a successful `/System/Info/Public` ping reported "online" even while the user's authenticated data calls were failing — and vice versa.
+
+### Time-window debounce (offline) + instant recovery (online)
+
+To stop the banner from flapping on a single dropped request, the transition to **offline** is debounced over a time window:
+
+- On the **first** `Network` failure, the monitor records `first_failure_at`.
+- It flips `is_server_reachable = false` only once `Network` failures have persisted continuously for `OFFLINE_CONFIRM_WINDOW` (5s) with no intervening success.
+- **Any** success (or server-answered error) clears `first_failure_at` and immediately marks reachable.
+
+Recovery is therefore instant and asymmetric: one good response brings the app back online, but a brief blip never trips the banner.
+
+### Offline-only recovery probe
 
 ```mermaid
 flowchart TB
-    Monitor["ConnectivityMonitor"] --> Poller["Background Task"]
-    Poller --> Check{"Server<br/>Reachable?"}
-    Check -->|"Yes"| Online["30s Interval"]
-    Check -->|"No"| Offline["5s Interval"]
-    Online --> Emit["Emit Events"]
-    Offline --> Emit
-    Emit --> Frontend["Frontend Store"]
+    Repo["OnlineRepository"] -->|"success / RepoError"| Monitor["ConnectivityMonitor"]
+    Monitor --> State{"is_server_reachable?"}
+    State -->|"Online"| NoProbe["No background polling<br/>(real traffic is the signal)"]
+    State -->|"Offline"| Probe["5s /System/Info/Public probe<br/>(recovery detector)"]
+    Probe -->|"reachable again"| Monitor
+    Monitor -->|"on change"| Emit["Emit connectivity:changed<br/>+ connectivity:reconnected"]
+    Emit --> Frontend["Frontend Store → banner"]
 ```
 
+While **online**, there is no background polling — real requests keep the state fresh. While **offline**, the fast 5s probe runs so an idle app still detects the server returning even when no user traffic is flowing.
+
 **Features:**
-- **Adaptive Polling**: 30s when online, 5s when offline (for quick reconnection detection)
-- **Event Emission**: Emits `connectivity:changed` and `connectivity:reconnected` events
-- **Manual Marking**: Can mark reachable/unreachable based on API call results
-- **Thread-Safe**: Uses Arc<RwLock<>> for shared state
+- **Traffic-driven**: Reachability follows the requests the user actually makes.
+- **Time-window debounce**: Offline declared only after `OFFLINE_CONFIRM_WINDOW` (5s) of sustained network failure; recovery is instant.
+- **Offline-only probe**: 5s `/System/Info/Public` probe runs only while offline.
+- **Event Emission**: Emits `connectivity:changed` and `connectivity:reconnected` events.
+- **Thread-Safe**: Uses `Arc<RwLock<>>` for shared state.
 
 **Tauri Commands:**
 | Command | Description |
 |---------|-------------|
-| `connectivity_check_server` | Manual reachability check |
+| `connectivity_check_server` | Manual reachability check (also used by the frontend's advisory `navigator.onLine` hint) |
 | `connectivity_set_server_url` | Update monitored server URL |
 | `connectivity_get_status` | Get current connectivity status |
-| `connectivity_start_monitoring` | Start background monitoring |
-| `connectivity_stop_monitoring` | Stop monitoring |
-| `connectivity_mark_reachable` | Mark server as reachable (after successful API call) |
-| `connectivity_mark_unreachable` | Mark server as unreachable (after failed API call) |
+| `connectivity_start_monitoring` | Start the offline recovery probe |
+| `connectivity_stop_monitoring` | Stop the probe |
+| `connectivity_mark_reachable` | Mark reachable — driven by `OnlineRepository` on every server success |
+| `connectivity_mark_unreachable` | Mark unreachable — driven by `OnlineRepository` on `RepoError::Network` (subject to debounce) |
 
 **Frontend Integration:**
 ```typescript
-// TypeScript store listens to Rust events
+// The store is a pure reflection of backend events — it no longer decides
+// reachability itself. navigator.onLine is advisory: it triggers an immediate
+// recheck rather than forcing the offline state.
 listen<{ isReachable: boolean }>("connectivity:changed", (event) => {
   updateConnectivityState(event.payload.isReachable);
 });
@@ -81,12 +114,14 @@ listen<{ isReachable: boolean }>("connectivity:changed", (event) => {
 The connectivity system provides resilience through multiple layers:
 
 1. **HTTP Client Layer**: Automatic retry with exponential backoff
-2. **Connectivity Monitoring**: Background reachability checks
-3. **Frontend Integration**: Offline mode detection and UI updates
+2. **Connectivity Monitoring**: Reachability derived from real repository traffic, with an offline-only recovery probe
+3. **Frontend Integration**: Offline mode detection and UI updates (a pure reflection of backend events)
 4. **Sync Queue**: Offline mutations queued for later (see [06-downloads-and-offline.md](06-downloads-and-offline.md))
 
 **Design Principles:**
-- **Fail Fast**: Don't retry 4xx errors (client errors, authentication)
-- **Fail Slow**: Retry network and 5xx errors with increasing delays
-- **Adaptive Polling**: Reduce polling frequency when online, increase when offline
-- **Event-Driven**: Frontend reacts to connectivity changes via events
+- **Single source of truth**: Reachability follows the outcome of real requests, classified via `RepoError`; the frontend store and the probe never compete to decide it.
+- **Fail Fast**: Don't retry 4xx errors (client errors, authentication).
+- **Fail Slow**: Retry network and 5xx errors with increasing delays.
+- **Debounced offline, instant online**: Declare offline only after a sustained failure window; recover on the first success.
+- **Probe only when needed**: Background polling runs only while offline, as a recovery detector.
+- **Event-Driven**: Frontend reacts to connectivity changes via events.

docs/architecture/README.md

@@ -15,6 +15,7 @@ JellyTau uses a client-server architecture: business logic lives in a comprehens
 - **Events + Polling hybrid**: Rust emits events the frontend listens to, and the UI also polls status on short intervals in a few hot spots (e.g. queue status in `library/+layout.svelte`, playback progress in `VideoPlayer.svelte`).
 - **Handle-Based Resources**: UUID handles for stateful Rust objects.
 - **Cache-First**: Parallel queries with intelligent fallback.
+- **Single source of truth for reachability**: Server reachability is derived from the outcome of *real repository traffic*, not a side-channel poller. The `OnlineRepository` reports each server result to the `ConnectivityMonitor` (classified via `RepoError`), which applies a time-window debounce before declaring the server offline and recovers instantly on the first success. The standalone `/System/Info/Public` probe runs *only while offline*, as a recovery detector for idle sessions.
 - **Poison-tolerant locking**: Shared `std::sync` state is accessed via the `MutexSafe`/`RwLockSafe` helpers in `utils/lock.rs`, which recover a poisoned lock instead of cascading a panic across the player.
 - **Graceful backend init**: If a native player backend (MPV/ExoPlayer) fails to initialize, the app falls back to a no-op backend and emits a `backend-init-failed` event rather than crashing.
 
@@ -79,9 +80,12 @@ flowchart TB
         Core --> Storage
         Repository --> HttpClient
         Repository --> DatabaseService
+        Repository -->|"reports server outcome<br/>(success / RepoError)"| ConnectivityMonitor
     end
 ```
 
+> The `Repository --> ConnectivityMonitor` edge is the source of truth for the offline/online banner: every server request the user actually makes updates reachability. The monitor's own polling is now an offline-only recovery probe (see [07-connectivity.md](07-connectivity.md)).
+
 ---
 
 ## Detailed Documentation
@@ -187,7 +191,7 @@ src/lib/
 
 **What moved to Rust (~3,500 lines of business logic):**
 1. **HTTP Client** (338 lines) - Retry logic with exponential backoff
-2. **Connectivity Monitor** (301 lines) - Adaptive polling, event emission
+2. **Connectivity Monitor** (301 lines) - Reachability derived from real repository traffic, time-window debounce, offline-only recovery probe, event emission
 3. **Repository Pattern** (1061 lines) - Cache-first hybrid with parallel racing
 4. **Database Service** - Async wrapper preventing UI freezing
 5. **Playback Mode** (303 lines) - Local/remote transfer coordination

src-tauri/src/commands/repository.rs

@@ -52,6 +52,7 @@ pub struct RepositoryManagerWrapper(pub RepositoryManager);
 pub async fn repository_create(
     manager: State<'_, RepositoryManagerWrapper>,
     db: State<'_, crate::commands::storage::DatabaseWrapper>,
+    connectivity: State<'_, crate::commands::connectivity::ConnectivityMonitorWrapper>,
     server_url: String,
     user_id: String,
     access_token: String,
@@ -68,9 +69,18 @@ pub async fn repository_create(
     })?;
     debug!("[REPO] HTTP client created successfully");
 
-    // Create online repository
+    // Grab a connectivity reporter so the online repository's server outcomes
+    // drive the reachability state the UI observes (source of truth for the
+    // offline/online banner). See docs/architecture/07-connectivity.md.
+    let connectivity_reporter = {
+        let monitor = connectivity.0.lock().await;
+        monitor.reporter()
+    };
+
+    // Create online repository wired to connectivity reporting
     debug!("[REPO] Creating online repository...");
-    let online = OnlineRepository::new(Arc::new(http_client), server_url, user_id.clone(), access_token);
+    let online = OnlineRepository::new(Arc::new(http_client), server_url, user_id.clone(), access_token)
+        .with_connectivity(connectivity_reporter);
     debug!("[REPO] Online repository created");
 
     // Create offline repository with async-safe database service

src-tauri/src/connectivity/mod.rs

@@ -1,16 +1,24 @@
 use std::sync::atomic::{AtomicBool, Ordering};
 use std::sync::Arc;
-use std::time::Duration;
+use std::time::{Duration, Instant};
 use tokio::sync::RwLock;
 use tauri::{AppHandle, Emitter};
 use serde::{Serialize, Deserialize};
 
 use crate::jellyfin::http_client::HttpClient;
 
-// Adaptive polling intervals (matches TypeScript)
-const AUTO_CHECK_INTERVAL_MS: u64 = 30000; // 30 seconds when online
+// Offline recovery probe interval.
+// Reachability while online is driven by real repository traffic, so there is
+// no online polling. While offline we probe quickly to detect the server
+// returning even when no user traffic is flowing.
 const RETRY_CHECK_INTERVAL_MS: u64 = 5000; // 5 seconds when offline
 
+// Time-window debounce for declaring the server offline.
+// A single dropped request must not trip the banner: we only flip to offline
+// once network failures have persisted continuously for this window with no
+// intervening success. Recovery (online) is instant on the first success.
+const OFFLINE_CONFIRM_WINDOW: Duration = Duration::from_secs(5);
+
 /// Connectivity status
 #[derive(specta::Type, Debug, Clone, Serialize, Deserialize)]
 #[serde(rename_all = "camelCase")]
@@ -45,210 +53,115 @@ struct ConnectivityChangeEvent {
     is_reachable: bool,
 }
 
-/// Connectivity monitor for tracking server reachability
-pub struct ConnectivityMonitor {
-    server_url: Arc<RwLock<Option<String>>>,
-    http_client: Arc<HttpClient>,
+/// Shared reachability state and transition logic.
+///
+/// This is the single place that mutates reachability and emits events. It is
+/// cheap to clone (all fields are `Arc`/`Option`) and is shared by:
+/// - the `ConnectivityMonitor` (commands, offline recovery probe), and
+/// - `OnlineRepository`, which reports the outcome of every server request.
+///
+/// Reachability is therefore driven by real traffic; the probe only fills the
+/// gap while offline.
+#[derive(Clone)]
+pub struct ConnectivityReporter {
     status: Arc<RwLock<ConnectivityStatus>>,
-    is_monitoring: Arc<AtomicBool>,
+    /// Timestamp of the first network failure in the current failure streak.
+    /// Used to debounce the transition to offline (see `OFFLINE_CONFIRM_WINDOW`).
+    first_failure_at: Arc<RwLock<Option<Instant>>>,
     app_handle: Option<AppHandle>,
 }
 
-impl ConnectivityMonitor {
-    /// Create a new connectivity monitor
-    pub fn new(http_client: HttpClient) -> Self {
+impl ConnectivityReporter {
+    fn new(status: Arc<RwLock<ConnectivityStatus>>, app_handle: Option<AppHandle>) -> Self {
         Self {
-            server_url: Arc::new(RwLock::new(None)),
-            http_client: Arc::new(http_client),
-            status: Arc::new(RwLock::new(ConnectivityStatus::default())),
-            is_monitoring: Arc::new(AtomicBool::new(false)),
-            app_handle: None,
+            status,
+            first_failure_at: Arc::new(RwLock::new(None)),
+            app_handle,
         }
     }
 
-    /// Set the Tauri app handle for event emission
-    pub fn set_app_handle(&mut self, app_handle: AppHandle) {
-        self.app_handle = Some(app_handle);
+    /// Current reachability as seen by this reporter (shared with the monitor
+    /// and the UI). Useful for callers that want to branch on connectivity.
+    pub async fn is_reachable(&self) -> bool {
+        self.status.read().await.is_server_reachable
     }
 
-    /// Update the server URL
-    pub async fn set_server_url(&self, url: String) {
-        log::info!("[ConnectivityMonitor] Setting server URL: {}", url);
-        let mut server_url = self.server_url.write().await;
-        *server_url = Some(url.clone());
-        drop(server_url);
-
-        // Check new server immediately
-        log::info!("[ConnectivityMonitor] Checking reachability of new server...");
-        let is_reachable = self.check_reachability().await;
-        log::info!("[ConnectivityMonitor] New server is {}", if is_reachable { "REACHABLE" } else { "UNREACHABLE" });
+    /// Test-only: force the reporter into the offline state without going through
+    /// the debounce, so other modules' tests can set up an "offline" precondition.
+    #[cfg(test)]
+    pub async fn mark_unreachable_for_test(&self) {
+        self.apply_probe_result(false, Some("forced offline (test)".to_string()))
+            .await;
     }
 
-    /// Get current connectivity status
-    pub async fn get_status(&self) -> ConnectivityStatus {
-        self.status.read().await.clone()
+    /// Report that a real server request succeeded (or that the server answered
+    /// at all, e.g. with 401/404/5xx). The server is up — recover instantly.
+    pub async fn report_success(&self) {
+        *self.first_failure_at.write().await = None;
+        self.set_reachable(true, None).await;
     }
 
-    /// Check if the Jellyfin server is reachable
-    pub async fn check_reachability(&self) -> bool {
-        // Mark as checking
-        {
-            let mut status = self.status.write().await;
-            status.is_checking = true;
+    /// Report that a real server request failed with a network-level error
+    /// (connection refused, timeout, DNS). Subject to the time-window debounce:
+    /// we only flip to offline once failures have persisted for
+    /// `OFFLINE_CONFIRM_WINDOW` with no intervening success.
+    pub async fn report_network_failure(&self, error: Option<String>) {
+        // If already offline, nothing to debounce.
+        if !self.status.read().await.is_server_reachable {
+            return;
         }
 
-        let server_url = self.server_url.read().await.clone();
-
-        if server_url.is_none() {
-            log::warn!("[ConnectivityMonitor] Cannot check reachability: No server URL configured");
-            let mut status = self.status.write().await;
-            status.is_server_reachable = false;
-            status.connection_error = Some("No server URL configured".to_string());
-            status.is_checking = false;
-            return false;
-        }
-
-        let url = server_url.unwrap();
-        let ping_url = format!("{}/System/Info/Public", url);
-
-        // Store previous reachability state
-        let was_reachable = {
-            let status = self.status.read().await;
-            status.is_server_reachable
+        let now = Instant::now();
+        let streak_start = {
+            let mut first = self.first_failure_at.write().await;
+            *first.get_or_insert(now)
         };
 
-        log::debug!("[ConnectivityMonitor] Pinging server: {}", ping_url);
-
-        // Attempt to ping the server
-        let is_reachable = self.http_client.ping(&ping_url).await;
-
-        log::debug!(
-            "[ConnectivityMonitor] Ping result: {} (was: {})",
-            if is_reachable { "SUCCESS" } else { "FAILED" },
-            if was_reachable { "reachable" } else { "unreachable" }
+        if now.duration_since(streak_start) >= OFFLINE_CONFIRM_WINDOW {
+            log::warn!(
+                "[ConnectivityMonitor] Network failures sustained for {:?}; declaring offline",
+                OFFLINE_CONFIRM_WINDOW
             );
+            self.set_reachable(false, error).await;
+        } else {
+            log::debug!(
+                "[ConnectivityMonitor] Network failure within debounce window; not yet offline"
+            );
+        }
+    }
 
-        // Update status
-        {
+    /// Apply a deliberate reachability probe result (offline recovery probe or a
+    /// manual check). Unlike `report_network_failure`, a probe is an explicit
+    /// reachability test, so its result is applied immediately without debounce.
+    async fn apply_probe_result(&self, is_reachable: bool, error: Option<String>) {
+        if is_reachable {
+            *self.first_failure_at.write().await = None;
+        }
+        self.set_reachable(is_reachable, error).await;
+    }
+
+    /// Core transition: update status and emit events only on an actual change.
+    async fn set_reachable(&self, is_reachable: bool, error: Option<String>) {
+        let was_reachable = {
             let mut status = self.status.write().await;
+            let was = status.is_server_reachable;
             status.is_server_reachable = is_reachable;
             status.last_checked = Some(chrono::Utc::now().to_rfc3339());
             status.connection_error = if is_reachable {
                 None
             } else {
-                Some("Server unreachable".to_string())
+                Some(error.unwrap_or_else(|| "Server unreachable".to_string()))
             };
             status.is_checking = false;
-        }
-
-        // Emit events if reachability changed
-        if is_reachable != was_reachable {
-            self.emit_connectivity_change(is_reachable).await;
-        }
-
-        // Emit reconnection event
-        if is_reachable && !was_reachable {
-            self.emit_server_reconnected().await;
-        }
-
-        is_reachable
-    }
-
-    /// Mark server as reachable (called after successful API call)
-    pub async fn mark_reachable(&self) {
-        let mut status = self.status.write().await;
-        let was_reachable = status.is_server_reachable;
-
-        status.is_server_reachable = true;
-        status.last_checked = Some(chrono::Utc::now().to_rfc3339());
-        status.connection_error = None;
-
-        drop(status);
-
-        if !was_reachable {
-            log::info!("[ConnectivityMonitor] Server marked as reachable (was unreachable)");
-            self.emit_connectivity_change(true).await;
-            self.emit_server_reconnected().await;
-        }
-    }
-
-    /// Mark server as unreachable (called after failed API call)
-    pub async fn mark_unreachable(&self, error: Option<String>) {
-        let mut status = self.status.write().await;
-        let was_reachable = status.is_server_reachable;
-
-        status.is_server_reachable = false;
-        status.last_checked = Some(chrono::Utc::now().to_rfc3339());
-        status.connection_error = error.or_else(|| Some("Server unreachable".to_string()));
-
-        let error_msg = status.connection_error.clone().unwrap_or_default();
-        drop(status);
-
-        if was_reachable {
-            log::warn!("[ConnectivityMonitor] Server marked as unreachable (was reachable): {}", error_msg);
-            self.emit_connectivity_change(false).await;
-        }
-    }
-
-    /// Start monitoring connectivity with adaptive polling
-    pub async fn start_monitoring(&self) {
-        if self.is_monitoring.swap(true, Ordering::SeqCst) {
-            log::info!("[ConnectivityMonitor] Already monitoring");
-            return;
-        }
-
-        log::info!("[ConnectivityMonitor] Starting connectivity monitoring");
-
-        // Perform immediate check before starting background task
-        // This ensures we get an accurate state right away instead of assuming offline
-        let is_reachable = self.check_reachability().await;
-        log::info!("[ConnectivityMonitor] Initial connectivity check: {}", if is_reachable { "ONLINE" } else { "OFFLINE" });
-
-        // Clone Arc references for the background task
-        let status = Arc::clone(&self.status);
-        let is_monitoring = Arc::clone(&self.is_monitoring);
-        let server_url = Arc::clone(&self.server_url);
-        let http_client = Arc::clone(&self.http_client);
-        let self_clone = Arc::new(ConnectivityMonitorHandle {
-            server_url,
-            http_client,
-            status,
-            app_handle: self.app_handle.clone(),
-        });
-
-        // Spawn background monitoring task
-        tokio::spawn(async move {
-            while is_monitoring.load(Ordering::SeqCst) {
-                // Determine interval based on current reachability
-                let interval_ms = {
-                    let status = self_clone.status.read().await;
-                    if status.is_server_reachable {
-                        AUTO_CHECK_INTERVAL_MS
-                    } else {
-                        RETRY_CHECK_INTERVAL_MS
-                    }
+            was
         };
 
-                // Wait for the interval
-                tokio::time::sleep(Duration::from_millis(interval_ms)).await;
-
-                // Check if still monitoring
-                if !is_monitoring.load(Ordering::SeqCst) {
-                    break;
+        if is_reachable != was_reachable {
+            self.emit_connectivity_change(is_reachable).await;
+            if is_reachable {
+                self.emit_server_reconnected().await;
             }
-
-                // Perform connectivity check
-                let _ = self_clone.check_reachability().await;
         }
-
-            log::info!("[ConnectivityMonitor] Stopped monitoring");
-        });
-    }
-
-    /// Stop monitoring connectivity
-    pub fn stop_monitoring(&self) {
-        log::info!("[ConnectivityMonitor] Stopping connectivity monitoring");
-        self.is_monitoring.store(false, Ordering::SeqCst);
     }
 
     /// Emit connectivity change event to frontend
@@ -275,74 +188,160 @@ impl ConnectivityMonitor {
     }
 }
 
-/// Handle for the background monitoring task
-struct ConnectivityMonitorHandle {
+/// Connectivity monitor for tracking server reachability.
+///
+/// Reachability is driven primarily by real repository traffic via the shared
+/// [`ConnectivityReporter`]. The monitor itself only runs an offline recovery
+/// probe (see `start_monitoring`) and serves the connectivity Tauri commands.
+pub struct ConnectivityMonitor {
     server_url: Arc<RwLock<Option<String>>>,
     http_client: Arc<HttpClient>,
-    status: Arc<RwLock<ConnectivityStatus>>,
-    app_handle: Option<AppHandle>,
+    reporter: ConnectivityReporter,
+    is_monitoring: Arc<AtomicBool>,
 }
 
-impl ConnectivityMonitorHandle {
-    async fn check_reachability(&self) -> bool {
+impl ConnectivityMonitor {
+    /// Create a new connectivity monitor
+    pub fn new(http_client: HttpClient) -> Self {
+        let status = Arc::new(RwLock::new(ConnectivityStatus::default()));
+        Self {
+            server_url: Arc::new(RwLock::new(None)),
+            http_client: Arc::new(http_client),
+            reporter: ConnectivityReporter::new(status, None),
+            is_monitoring: Arc::new(AtomicBool::new(false)),
+        }
+    }
+
+    /// Set the Tauri app handle for event emission.
+    /// Must be called before the reporter is shared with the repository.
+    pub fn set_app_handle(&mut self, app_handle: AppHandle) {
+        self.reporter.app_handle = Some(app_handle);
+    }
+
+    /// Get a cheap, cloneable reporter so the repository can feed server
+    /// outcomes into the same reachability state the UI observes.
+    pub fn reporter(&self) -> ConnectivityReporter {
+        self.reporter.clone()
+    }
+
+    /// Update the server URL
+    pub async fn set_server_url(&self, url: String) {
+        log::info!("[ConnectivityMonitor] Setting server URL: {}", url);
+        *self.server_url.write().await = Some(url);
+
+        // Check new server immediately
+        log::info!("[ConnectivityMonitor] Checking reachability of new server...");
+        let is_reachable = self.check_reachability().await;
+        log::info!("[ConnectivityMonitor] New server is {}", if is_reachable { "REACHABLE" } else { "UNREACHABLE" });
+    }
+
+    /// Get current connectivity status
+    pub async fn get_status(&self) -> ConnectivityStatus {
+        self.reporter.status.read().await.clone()
+    }
+
+    /// Deliberately probe the server's reachability (manual check / recovery probe).
+    /// The result is applied immediately (no debounce) since this is an explicit test.
+    pub async fn check_reachability(&self) -> bool {
+        {
+            let mut status = self.reporter.status.write().await;
+            status.is_checking = true;
+        }
+
         let server_url = self.server_url.read().await.clone();
 
-        if server_url.is_none() {
+        let Some(url) = server_url else {
+            log::warn!("[ConnectivityMonitor] Cannot check reachability: No server URL configured");
+            self.reporter
+                .apply_probe_result(false, Some("No server URL configured".to_string()))
+                .await;
             return false;
-        }
+        };
 
-        let url = server_url.unwrap();
         let ping_url = format!("{}/System/Info/Public", url);
+        log::debug!("[ConnectivityMonitor] Pinging server: {}", ping_url);
 
-        // Store previous reachability state
-        let was_reachable = {
-            let status = self.status.read().await;
-            status.is_server_reachable
-        };
-
-        // Attempt to ping the server
         let is_reachable = self.http_client.ping(&ping_url).await;
+        log::debug!(
+            "[ConnectivityMonitor] Ping result: {}",
+            if is_reachable { "SUCCESS" } else { "FAILED" }
+        );
 
-        // Update status
-        {
-            let mut status = self.status.write().await;
-            status.is_server_reachable = is_reachable;
-            status.last_checked = Some(chrono::Utc::now().to_rfc3339());
-            status.connection_error = if is_reachable {
-                None
-            } else {
-                Some("Server unreachable".to_string())
-            };
-        }
-
-        // Emit events if reachability changed
-        if is_reachable != was_reachable {
-            self.emit_connectivity_change(is_reachable).await;
-        }
-
-        // Emit reconnection event
-        if is_reachable && !was_reachable {
-            self.emit_server_reconnected().await;
-        }
-
+        self.reporter.apply_probe_result(is_reachable, None).await;
         is_reachable
     }
 
-    async fn emit_connectivity_change(&self, is_reachable: bool) {
-        if let Some(app_handle) = &self.app_handle {
-            let event = ConnectivityChangeEvent { is_reachable };
-            if let Err(e) = app_handle.emit("connectivity:changed", event) {
-                log::error!("[ConnectivityMonitor] Failed to emit connectivity change event: {}", e);
+    /// Mark server as reachable (called after successful API call / login)
+    pub async fn mark_reachable(&self) {
+        self.reporter.report_success().await;
     }
+
+    /// Mark server as unreachable directly.
+    ///
+    /// Used by deliberate signals (e.g. a failed login/connect) where the caller
+    /// knows the server is unreachable now. Repository traffic should prefer
+    /// `reporter().report_network_failure()` so the debounce applies.
+    pub async fn mark_unreachable(&self, error: Option<String>) {
+        self.reporter.apply_probe_result(false, error).await;
+    }
+
+    /// Start the offline recovery probe.
+    ///
+    /// While **online**, reachability is kept fresh by real traffic, so the probe
+    /// idles. While **offline**, it polls `/System/Info/Public` every
+    /// `RETRY_CHECK_INTERVAL_MS` to detect the server returning even when no user
+    /// traffic is flowing.
+    pub async fn start_monitoring(&self) {
+        if self.is_monitoring.swap(true, Ordering::SeqCst) {
+            log::info!("[ConnectivityMonitor] Already monitoring");
+            return;
+        }
+
+        log::info!("[ConnectivityMonitor] Starting connectivity monitoring (offline recovery probe)");
+
+        // Perform an immediate check so startup reflects reality quickly.
+        let is_reachable = self.check_reachability().await;
+        log::info!("[ConnectivityMonitor] Initial connectivity check: {}", if is_reachable { "ONLINE" } else { "OFFLINE" });
+
+        let is_monitoring = Arc::clone(&self.is_monitoring);
+        let server_url = Arc::clone(&self.server_url);
+        let http_client = Arc::clone(&self.http_client);
+        let reporter = self.reporter.clone();
+
+        tokio::spawn(async move {
+            while is_monitoring.load(Ordering::SeqCst) {
+                tokio::time::sleep(Duration::from_millis(RETRY_CHECK_INTERVAL_MS)).await;
+
+                if !is_monitoring.load(Ordering::SeqCst) {
+                    break;
+                }
+
+                // Only probe while offline — real traffic is the signal when online.
+                if reporter.status.read().await.is_server_reachable {
+                    continue;
+                }
+
+                let Some(url) = server_url.read().await.clone() else {
+                    continue;
+                };
+                let ping_url = format!("{}/System/Info/Public", url);
+                let is_reachable = http_client.ping(&ping_url).await;
+
+                // Probe only ever recovers us to online; a failed probe leaves us
+                // offline without re-emitting (no change).
+                if is_reachable {
+                    reporter.apply_probe_result(true, None).await;
                 }
             }
 
-    async fn emit_server_reconnected(&self) {
-        if let Some(app_handle) = &self.app_handle {
-            if let Err(e) = app_handle.emit("connectivity:reconnected", ()) {
-                log::error!("[ConnectivityMonitor] Failed to emit reconnection event: {}", e);
-            }
+            log::info!("[ConnectivityMonitor] Stopped monitoring");
+        });
     }
+
+    /// Stop monitoring connectivity
+    pub fn stop_monitoring(&self) {
+        log::info!("[ConnectivityMonitor] Stopping connectivity monitoring");
+        self.is_monitoring.store(false, Ordering::SeqCst);
     }
 }
 
@@ -350,11 +349,22 @@ impl ConnectivityMonitorHandle {
 mod tests {
     use super::*;
 
+    /// Build a reporter backed by a fresh (optimistic) status, with no app handle.
+    /// Event emission is a no-op without a handle, which is exactly what we want
+    /// for unit-testing the reachability state transitions.
+    fn test_reporter() -> ConnectivityReporter {
+        ConnectivityReporter::new(Arc::new(RwLock::new(ConnectivityStatus::default())), None)
+    }
+
+    async fn is_reachable(reporter: &ConnectivityReporter) -> bool {
+        reporter.status.read().await.is_server_reachable
+    }
+
     #[test]
     fn test_intervals() {
-        // Verify intervals match TypeScript
-        assert_eq!(AUTO_CHECK_INTERVAL_MS, 30000);
+        // Offline recovery probe interval (online has no polling).
         assert_eq!(RETRY_CHECK_INTERVAL_MS, 5000);
+        assert_eq!(OFFLINE_CONFIRM_WINDOW, Duration::from_secs(5));
     }
 
     #[tokio::test]
@@ -366,4 +376,120 @@ mod tests {
         assert!(status.connection_error.is_none());
         assert!(!status.is_checking);
     }
+
+    /// A single (or brief) network failure must NOT flip the app offline:
+    /// the time-window debounce keeps us online until the failure persists.
+    ///
+    /// @req-test: UR-002 - Access media when online or offline
+    #[tokio::test]
+    async fn test_single_network_failure_does_not_go_offline() {
+        let reporter = test_reporter();
+        assert!(is_reachable(&reporter).await, "starts online");
+
+        reporter
+            .report_network_failure(Some("timeout".to_string()))
+            .await;
+
+        assert!(
+            is_reachable(&reporter).await,
+            "one network failure within the debounce window stays online"
+        );
+        // But the failure streak is now being tracked.
+        assert!(reporter.first_failure_at.read().await.is_some());
+    }
+
+    /// Once failures persist past OFFLINE_CONFIRM_WINDOW, we flip offline.
+    /// We simulate elapsed time by backdating the streak start.
+    ///
+    /// @req-test: UR-002 - Access media when online or offline
+    #[tokio::test]
+    async fn test_sustained_network_failure_goes_offline() {
+        let reporter = test_reporter();
+
+        // First failure starts the streak.
+        reporter.report_network_failure(None).await;
+        assert!(is_reachable(&reporter).await);
+
+        // Backdate the streak start to before the window.
+        {
+            let mut first = reporter.first_failure_at.write().await;
+            *first = Some(Instant::now() - OFFLINE_CONFIRM_WINDOW - Duration::from_secs(1));
+        }
+
+        // Next failure now exceeds the window → offline.
+        reporter
+            .report_network_failure(Some("connection refused".to_string()))
+            .await;
+        assert!(
+            !is_reachable(&reporter).await,
+            "sustained network failure flips to offline"
+        );
+    }
+
+    /// A success during a failure streak clears the streak and keeps us online —
+    /// recovery is instant and never trips the banner.
+    #[tokio::test]
+    async fn test_success_clears_failure_streak() {
+        let reporter = test_reporter();
+
+        reporter.report_network_failure(None).await;
+        assert!(reporter.first_failure_at.read().await.is_some());
+
+        reporter.report_success().await;
+
+        assert!(is_reachable(&reporter).await);
+        assert!(
+            reporter.first_failure_at.read().await.is_none(),
+            "success resets the debounce streak"
+        );
+    }
+
+    /// First success after being offline recovers instantly (no debounce on the
+    /// way back up).
+    #[tokio::test]
+    async fn test_recovery_is_instant() {
+        let reporter = test_reporter();
+
+        // Force offline.
+        reporter.apply_probe_result(false, Some("down".to_string())).await;
+        assert!(!is_reachable(&reporter).await);
+
+        // A single success brings us straight back online.
+        reporter.report_success().await;
+        assert!(is_reachable(&reporter).await);
+        let status = reporter.status.read().await;
+        assert!(status.connection_error.is_none());
+    }
+
+    /// Server-answered errors (401/404/5xx) are reported via report_success
+    /// by the repository, because the server is demonstrably reachable. This
+    /// test documents that contract: report_success means "server is up".
+    #[tokio::test]
+    async fn test_server_answered_error_counts_as_reachable() {
+        let reporter = test_reporter();
+
+        // Simulate being offline, then the server answers (even with an error).
+        reporter.apply_probe_result(false, None).await;
+        assert!(!is_reachable(&reporter).await);
+
+        // Repository maps Authentication/NotFound/Server errors to report_success.
+        reporter.report_success().await;
+        assert!(
+            is_reachable(&reporter).await,
+            "a server that answers (even with 4xx/5xx) is reachable"
+        );
+    }
+
+    /// report_network_failure is a no-op once already offline (nothing to debounce,
+    /// no duplicate events).
+    #[tokio::test]
+    async fn test_network_failure_noop_when_already_offline() {
+        let reporter = test_reporter();
+        reporter.apply_probe_result(false, None).await;
+        assert!(!is_reachable(&reporter).await);
+
+        // Should not panic or change state.
+        reporter.report_network_failure(Some("still down".to_string())).await;
+        assert!(!is_reachable(&reporter).await);
+    }
 }

src-tauri/src/repository/online.rs

@@ -7,6 +7,7 @@ use log::{debug, error, info};
 use log::warn;
 use serde::{Deserialize, Serialize};
 
+use crate::connectivity::ConnectivityReporter;
 use crate::jellyfin::HttpClient;
 use super::{MediaRepository, types::*};
 
@@ -16,6 +17,10 @@ pub struct OnlineRepository {
     server_url: String,
     user_id: String,
     access_token: String,
+    /// Reports the outcome of every server request to the connectivity monitor.
+    /// This is the source of truth for the offline/online banner. `None` in
+    /// tests / contexts where connectivity tracking isn't wired up.
+    connectivity: Option<ConnectivityReporter>,
 }
 
 impl OnlineRepository {
@@ -30,6 +35,44 @@ impl OnlineRepository {
             server_url,
             user_id,
             access_token,
+            connectivity: None,
+        }
+    }
+
+    /// Attach a connectivity reporter so server outcomes drive the reachability
+    /// state observed by the UI. See `report_outcome`.
+    pub fn with_connectivity(mut self, reporter: ConnectivityReporter) -> Self {
+        self.connectivity = Some(reporter);
+        self
+    }
+
+    /// Feed a request outcome into the connectivity monitor.
+    ///
+    /// Classification (matches docs/architecture/07-connectivity.md):
+    /// - `Ok` / `Authentication` / `NotFound` / `Server` → the server answered,
+    ///   so it is reachable → `report_success` (instant recovery).
+    /// - `Network` → connection-level failure → `report_network_failure`
+    ///   (subject to the time-window debounce before going offline).
+    /// - `Database` → not a server signal → ignored.
+    async fn report_outcome<T>(&self, result: &Result<T, RepoError>) {
+        let Some(reporter) = &self.connectivity else {
+            return;
+        };
+
+        match result {
+            Ok(_)
+            | Err(RepoError::Authentication { .. })
+            | Err(RepoError::NotFound { .. })
+            | Err(RepoError::Server { .. }) => {
+                reporter.report_success().await;
+            }
+            Err(RepoError::Network { message }) => {
+                reporter.report_network_failure(Some(message.clone())).await;
+            }
+            Err(RepoError::Database { .. }) | Err(RepoError::Offline) => {
+                // Local-side errors (cache failure / already-offline) — not a
+                // statement about the server's reachability, so ignore them.
+            }
         }
     }
 
@@ -63,6 +106,12 @@ impl OnlineRepository {
 
     /// Make authenticated GET request
     async fn get_json<T: for<'de> Deserialize<'de>>(&self, endpoint: &str) -> Result<T, RepoError> {
+        let result = self.get_json_inner(endpoint).await;
+        self.report_outcome(&result).await;
+        result
+    }
+
+    async fn get_json_inner<T: for<'de> Deserialize<'de>>(&self, endpoint: &str) -> Result<T, RepoError> {
         let url = format!("{}{}", self.server_url, endpoint);
 
         let request = self.http_client.client.get(&url)
@@ -110,6 +159,12 @@ impl OnlineRepository {
 
     /// Make authenticated POST request
     async fn post_json<T: Serialize>(&self, endpoint: &str, body: &T) -> Result<(), RepoError> {
+        let result = self.post_json_inner(endpoint, body).await;
+        self.report_outcome(&result).await;
+        result
+    }
+
+    async fn post_json_inner<T: Serialize>(&self, endpoint: &str, body: &T) -> Result<(), RepoError> {
         let url = format!("{}{}", self.server_url, endpoint);
 
         let request = self.http_client.client.post(&url)
@@ -145,6 +200,16 @@ impl OnlineRepository {
         &self,
         endpoint: &str,
         body: &T,
+    ) -> Result<R, RepoError> {
+        let result = self.post_json_response_inner(endpoint, body).await;
+        self.report_outcome(&result).await;
+        result
+    }
+
+    async fn post_json_response_inner<T: Serialize, R: for<'de> Deserialize<'de>>(
+        &self,
+        endpoint: &str,
+        body: &T,
     ) -> Result<R, RepoError> {
         let url = format!("{}{}", self.server_url, endpoint);
 
@@ -1156,6 +1221,7 @@ impl MediaRepository for OnlineRepository {
         let endpoint = format!("/Users/{}/FavoriteItems/{}", self.user_id, item_id);
         let url = format!("{}{}", self.server_url, endpoint);
 
+        let result = async {
             let request = self.http_client.client.delete(&url)
                 .header("X-Emby-Authorization", self.auth_header())
                 .build()
@@ -1174,6 +1240,11 @@ impl MediaRepository for OnlineRepository {
 
             Ok(())
         }
+        .await;
+
+        self.report_outcome(&result).await;
+        result
+    }
 
     async fn get_person(&self, person_id: &str) -> Result<MediaItem, RepoError> {
         let endpoint = format!("/Users/{}/Items/{}", self.user_id, person_id);
@@ -1397,6 +1468,91 @@ mod tests {
         )
     }
 
+    /// Build a repository wired to a real ConnectivityReporter so we can assert
+    /// how `report_outcome` classifies each `RepoError` into reachability.
+    /// (No app handle → event emission is a harmless no-op.)
+    fn create_test_repository_with_connectivity(
+    ) -> (OnlineRepository, crate::connectivity::ConnectivityReporter) {
+        let monitor_http = HttpClient::new(crate::jellyfin::HttpConfig::default())
+            .expect("Failed to create HTTP client for monitor");
+        let monitor = crate::connectivity::ConnectivityMonitor::new(monitor_http);
+        let reporter = monitor.reporter();
+        let repo = create_test_repository().with_connectivity(reporter.clone());
+        (repo, reporter)
+    }
+
+    /// `report_outcome` is the seam between repository traffic and the
+    /// connectivity monitor. Verify each `RepoError` variant routes correctly:
+    /// - the server answering at all (Ok / 401 / 404 / 5xx) ⇒ reachable
+    /// - a network-level failure ⇒ marked unreachable (debounce reduced for test)
+    /// - local-side errors (Database / Offline) ⇒ no effect on reachability
+    ///
+    /// @req-test: UR-002 - Access media when online or offline
+    /// @req-test: DR-013 - Repository pattern for online/offline data access
+    #[tokio::test]
+    async fn test_report_outcome_classifies_server_answered_as_reachable() {
+        let (repo, reporter) = create_test_repository_with_connectivity();
+
+        // Drive offline first so we can observe "recover to reachable".
+        for err in [
+            RepoError::Authentication { message: "401".into() },
+            RepoError::NotFound { message: "404".into() },
+            RepoError::Server { message: "500".into() },
+        ] {
+            reporter.mark_unreachable_for_test().await;
+            assert!(!reporter.is_reachable().await, "precondition: offline");
+
+            let result: Result<(), RepoError> = Err(err);
+            repo.report_outcome(&result).await;
+
+            assert!(
+                reporter.is_reachable().await,
+                "a server that answers should be reported reachable"
+            );
+        }
+
+        // Ok should also report reachable.
+        reporter.mark_unreachable_for_test().await;
+        let ok: Result<(), RepoError> = Ok(());
+        repo.report_outcome(&ok).await;
+        assert!(reporter.is_reachable().await, "Ok ⇒ reachable");
+    }
+
+    /// Local-side errors must NOT flip reachability — they say nothing about the
+    /// server.
+    #[tokio::test]
+    async fn test_report_outcome_ignores_local_errors() {
+        let (repo, reporter) = create_test_repository_with_connectivity();
+
+        // Force offline, then a Database/Offline error must leave it offline
+        // (not falsely report reachable).
+        reporter.mark_unreachable_for_test().await;
+        for err in [RepoError::Database { message: "cache".into() }, RepoError::Offline] {
+            let result: Result<(), RepoError> = Err(err);
+            repo.report_outcome(&result).await;
+            assert!(
+                !reporter.is_reachable().await,
+                "local-side error must not change reachability"
+            );
+        }
+    }
+
+    /// A network error routes through the debounced path. A single failure stays
+    /// online (debounce window not yet elapsed).
+    #[tokio::test]
+    async fn test_report_outcome_network_error_is_debounced() {
+        let (repo, reporter) = create_test_repository_with_connectivity();
+        assert!(reporter.is_reachable().await, "starts online");
+
+        let result: Result<(), RepoError> = Err(RepoError::Network { message: "timeout".into() });
+        repo.report_outcome(&result).await;
+
+        assert!(
+            reporter.is_reachable().await,
+            "a single network failure stays online (debounced)"
+        );
+    }
+
     #[tokio::test]
     async fn test_get_audio_stream_url_formats_correctly() {
         let repo = create_test_repository();

src/lib/stores/connectivity.ts

@@ -1,7 +1,12 @@
 // Connectivity state store for offline support
 //
-// Simplified wrapper over Rust connectivity monitor.
-// The Rust backend handles all polling, reachability checks, and adaptive intervals.
+// Pure reflection of the Rust ConnectivityMonitor. Reachability is the single
+// source of truth in Rust, derived from real repository traffic (success/
+// RepoError), with a time-window debounce before going offline and instant
+// recovery. This store only listens for `connectivity:changed` events and
+// mirrors status; it does not decide reachability itself. navigator.onLine is
+// advisory and triggers a recheck rather than forcing offline.
+// See docs/architecture/07-connectivity.md.
 // TRACES: UR-002 | DR-013
 
 import { writable, derived } from "svelte/store";
@@ -61,22 +66,27 @@ function createConnectivityStore() {
       }
     });
 
-    // Listen to browser online/offline events and update state
+    // Listen to browser online/offline events. These are ADVISORY only — the
+    // Rust ConnectivityMonitor (fed by real repository traffic) is the source of
+    // truth for server reachability. navigator.onLine can be wrong (e.g. a
+    // LAN-only server is still reachable while the browser reports "offline"),
+    // so we use these events to trigger an immediate recheck rather than forcing
+    // the offline state. See docs/architecture/07-connectivity.md.
     window.addEventListener("online", () => {
       update((s) => ({ ...s, isOnline: true }));
+      // Device regained network — ask the backend to re-verify the server now.
+      checkServerReachable().catch((err) => {
+        console.debug("[ConnectivityStore] Recheck after 'online' failed:", err);
+      });
     });
 
     window.addEventListener("offline", () => {
-      update((s) => ({
-        ...s,
-        isOnline: false,
-        isServerReachable: false,
-        connectionError: "Device is offline",
-      }));
-
-      if (eventHandlers.onConnectivityChange) {
-        eventHandlers.onConnectivityChange(false);
-      }
+      // Reflect the browser's view of the device link, but let the backend
+      // decide whether the server is actually reachable.
+      update((s) => ({ ...s, isOnline: false }));
+      checkServerReachable().catch((err) => {
+        console.debug("[ConnectivityStore] Recheck after 'offline' failed:", err);
+      });
     });
   }
 
@@ -199,43 +209,11 @@ function createConnectivityStore() {
     return checkServerReachable();
   }
 
-  /**
-   * Mark server as reachable (e.g., after successful API call)
-   */
-  async function markReachable(): Promise<void> {
-    try {
-      await commands.connectivityMarkReachable();
-
-      // Update local state
-      update((s) => ({
-        ...s,
-        isServerReachable: true,
-        lastChecked: new Date(),
-        connectionError: null,
-      }));
-    } catch (error) {
-      console.error("[ConnectivityStore] Failed to mark reachable:", error);
-    }
-  }
-
-  /**
-   * Mark server as unreachable (e.g., after failed API call)
-   */
-  async function markUnreachable(error?: string): Promise<void> {
-    try {
-      await commands.connectivityMarkUnreachable(error ?? null);
-
-      // Update local state
-      update((s) => ({
-        ...s,
-        isServerReachable: false,
-        lastChecked: new Date(),
-        connectionError: error || "Server unreachable",
-      }));
-    } catch (err) {
-      console.error("[ConnectivityStore] Failed to mark unreachable:", err);
-    }
-  }
+  // Reachability is now driven by the Rust backend from real repository
+  // traffic (see docs/architecture/07-connectivity.md). The frontend no longer
+  // mutates reachability itself, so the former markReachable/markUnreachable
+  // helpers have been removed. The underlying Rust commands remain available
+  // for deliberate signals if ever needed.
 
   return {
     subscribe,
@@ -244,8 +222,6 @@ function createConnectivityStore() {
     setServerUrl,
     forceCheck,
     checkServerReachable,
-    markReachable,
-    markUnreachable,
     isNetworkError,
   };
 }