diff --git a/docs/architecture/flows/event-bus-architecture.puml b/docs/architecture/flows/event-bus-architecture.puml
new file mode 100644
index 0000000..afaf1c3
--- /dev/null
+++ b/docs/architecture/flows/event-bus-architecture.puml
@@ -0,0 +1,86 @@
+@startuml Event Bus Architecture
+skinparam componentAlign center
+title Event Bus: In-Process Pub/Sub Architecture
+
+package "Publishers" {
+    [Workflow Goroutine 1\n(album A, LOSSLESS)] as WF1
+    [Workflow Goroutine 2\n(album B, LOSSY)] as WF2
+}
+
+database "PostgreSQL" as DB {
+    [workflow_runs] as WR
+    [album_events] as AE
+}
+
+package "Event Bus (in-memory)" {
+    [Topic: albumA:LOSSLESS] as T1
+    [Topic: albumB:LOSSY] as T2
+    [Global Subscribers] as GS
+}
+
+package "Subscribers" {
+    [MonitorAlbumStream\nClient A (album A)] as S1
+    [MonitorAlbumStream\nClient B (album A)] as S2
+    [SubscribeEvents\nClient C (global)] as S3
+}
+
+WF1 --> DB : 1. Write event\n(synchronous)
+WF1 --> T1 : 2. Publish\n(async notification)
+
+WF2 --> DB : 1. Write event
+WF2 --> T2 : 2. Publish
+
+T1 --> S1 : Ring buffer\n(per subscriber)
+T1 --> S2 : Ring buffer
+T1 --> GS
+T2 --> GS
+
+GS --> S3 : Ring buffer
+
+note right of DB
+  **Source of truth.**
+  Events survive restarts.
+  Replay via seq numbers.
+end note
+
+note right of T1
+  **Ephemeral notification.**
+  Ring buffer per subscriber.
+  Slow subscribers: overwrite oldest.
+  No backpressure on publishers.
+end note
+
+note bottom of S1
+  Client disconnect removes
+  subscriber from topic.
+  Workflow continues.
+end note
+
+== Subscription Lifecycle ==
+
+note as N1
+**Subscribe flow:**
+1. Client calls MonitorAlbumStream or SubscribeEvents
+2. Server subscribes to EventBus (per-topic or global)
+3. Server queries DB for historical events (replay)
+4. Server bridges: EventBus → gRPC stream
+5. On disconnect: cleanup func unsubscribes
+
+**Topic cleanup:**
+When last subscriber leaves AND workflow completed:
+topic removed from EventBus map.
+end note
+
+== Recovery on Restart ==
+
+note as N2
+**Server restart recovery:**
+1. Query workflow_runs WHERE status = 'running'
+2. For each stale run:
+   - If active download exists → mark completed
+   - Otherwise → mark failed ("server restarted")
+3. RecoverOrphanedDownloads reschedules poll jobs
+4. New workflows start fresh (no goroutine resurrection)
+end note
+
+@enduml
diff --git a/docs/architecture/flows/monitor-album-stream-already-owned.puml b/docs/architecture/flows/monitor-album-stream-already-owned.puml
index 34c3bbf..4d4400a 100644
--- a/docs/architecture/flows/monitor-album-stream-already-owned.puml
+++ b/docs/architecture/flows/monitor-album-stream-already-owned.puml
@@ -5,9 +5,16 @@ title MonitorAlbumStream: Already Owned Handling
 
 actor Client
 participant "monitorWorkflow" as Workflow
+participant "EventBus" as Bus
 database "PostgreSQL" as DB
 participant "IndexerService" as Indexer
 
+note over Client, Indexer #lightblue
+All events are persisted to album_events table (DB first)
+then published to EventBus for live subscribers.
+In automatic mode, workflow runs as background goroutine.
+end note
+
 == Scenario A: Automatic Mode - Early Return ==
 
 Client -> Workflow: StartMonitorRequest(mode=AUTOMATIC)
diff --git a/docs/architecture/flows/monitor-album-stream-automatic.puml b/docs/architecture/flows/monitor-album-stream-automatic.puml
index b029b28..f834957 100644
--- a/docs/architecture/flows/monitor-album-stream-automatic.puml
+++ b/docs/architecture/flows/monitor-album-stream-automatic.puml
@@ -1,113 +1,99 @@
 @startuml MonitorAlbumStream - Automatic Mode Happy Path
 skinparam sequenceMessageAlign center
 skinparam responseMessageBelowArrow true
-title MonitorAlbumStream: Automatic Mode (Status Updates Only)
+title MonitorAlbumStream: Automatic Mode (Fire-and-Forget)
 
 actor Client
 participant "gRPC Server" as Server
-participant "monitorWorkflow" as Workflow
-participant "MusicAgregatorService" as Service
-participant "MetadataService" as Metadata
-database "metadata-agregator\n(gRPC)" as MetaGRPC
+participant "WorkflowRegistry" as Registry
+participant "EventBus" as Bus
+participant "monitorWorkflow\n(background goroutine)" as Workflow
 database "PostgreSQL" as DB
-participant "IndexerService\n(Jackett)" as Indexer
+participant "MusicAgregatorService" as Service
+participant "IndexerService" as Indexer
 participant "MagnetResolver" as Magnet
 participant "TorrentClient\n(qBittorrent)" as QBit
-participant "River Queue" as River
 
-== 1. Initialize Bidirectional Stream ==
+== 1. Initialize Stream ==
 
 Client -> Server: MonitorAlbumStream()
-note right: Opens bidirectional\ngRPC stream
 Server -> Client: stream established
 
 Client -> Server: StartMonitorRequest
-note right: album_id, quality, tracker\nmode = AUTOMATIC
+note right: album_id, quality\nmode = AUTOMATIC
 
-Server -> Workflow: newMonitorWorkflow(stream, req, service)
-note right: Creates workflow with\ndecisions channel and\nreceiver goroutine
+== 2. Start or Subscribe to Workflow ==
 
-== 2. Fetch Album Metadata ==
+Server -> Registry: GetOrCreate(albumID, quality)
 
-Workflow ->> Client: StatusUpdate(FETCHING_METADATA)
-note right #lightblue: "Fetching metadata..."
+alt New workflow
+    Registry --> Server: (entry, created=true)
 
-Workflow -> Service: getAlbumWithPersist(ctx, album_id)
-Service -> Metadata: GetAlbum(album_id)
-Metadata -> MetaGRPC: GetAlbum(id)
-MetaGRPC --> Metadata: Album
-Metadata -> DB: artists.Create / albums.Create
-Metadata --> Service: Album
+    Server -> Bus: Subscribe(topic)
+    note right #lightblue: Subscribe BEFORE\nstarting goroutine\n(no missed events)
 
-Workflow ->> Client: StatusUpdate(FETCHING_METADATA)
-note right #lightblue: "Got: Artist - Title"\nData: StreamAlbumInfo
+    Server -> Workflow: go workflow.run(entry.Ctx)
+    note right #lightyellow: Background goroutine.\nDecoupled from stream context.\nClient can disconnect freely.
+else Existing workflow
+    Registry --> Server: (entry, created=false)
 
-== 3. Check If Already Owned ==
+    Server -> Bus: Subscribe(topic)
 
-Workflow ->> Client: StatusUpdate(CHECKING_OWNED)
-note right #lightblue: "Checking ownership..."
-
-Workflow -> DB: downloads.HasAlbumInQuality()
-DB --> Workflow: false
-
-== 4. Search Indexers ==
-
-Workflow ->> Client: StatusUpdate(SEARCHING_INDEXER)
-note right #lightblue: "Searching indexers..."
-
-Workflow -> Indexer: Search(query, tracker)
-Indexer --> Workflow: SearchResponse (N items)
-
-== 5. Parse Releases ==
-
-loop for each search result
-    Workflow -> Magnet: Resolve(magnet_uri)
-    Magnet --> Workflow: torrent metadata
-    Workflow -> Workflow: ParseTorrent()
+    Server -> DB: Replay album_events\n(since last seq)
+    DB --> Server: historical events
+    Server ->> Client: replayed events
 end
 
-Workflow ->> Client: StatusUpdate(PARSING_RESULTS)
-note right #lightblue: "Parsed M from N torrents"\nData: TorrentList
+== 3. Event Bridge (concurrent) ==
 
-== 6. Filter by Quality ==
+note over Client, Workflow #lightblue
+**Left side**: Event bus → gRPC stream bridge
+**Right side**: Workflow executing in background
+Both run concurrently. Client disconnect only stops the bridge.
+end note
 
-Workflow ->> Client: StatusUpdate(FILTERING_QUALITY)
-note right #lightblue: "Filtering by quality..."
+|||
 
-Workflow -> Workflow: filterByQuality(parsed, quality)
+Workflow -> DB: album_events.Create(FETCHING_METADATA)
+Workflow -> Bus: Publish(FETCHING_METADATA)
+Bus ->> Server: event notification
+Server ->> Client: StatusUpdate(FETCHING_METADATA)
 
-== 7. Select Best Release ==
+Workflow -> Service: getAlbumWithPersist()
+Workflow -> DB: workflow_runs.Create(albumID, quality)
 
-Workflow -> Workflow: selectBestRelease(filtered)
-note right: Highest seeder count wins\n(automatic selection)
+Workflow -> DB: album_events.Create(CHECKING_OWNED)
+Workflow -> Bus: Publish(CHECKING_OWNED)
+Bus ->> Server: event notification
+Server ->> Client: StatusUpdate(CHECKING_OWNED)
 
-Workflow ->> Client: StatusUpdate(SELECTING_RELEASE)
-note right #lightblue: "Selected: Title (N seeders)"\nData: ReleaseInfo
+Workflow -> Indexer: Search()
+Workflow -> DB: album_events.Create(PARSING_RESULTS)
+Workflow -> Bus: Publish(PARSING_RESULTS)
+Bus ->> Server: event notification
+Server ->> Client: StatusUpdate(PARSING_RESULTS)
 
-== 8. Add to Torrent Client ==
+Workflow -> Workflow: filterByQuality + selectBest
 
-Workflow ->> Client: StatusUpdate(ADDING_TORRENT)
-note right #lightblue: "Adding torrent..."
+Workflow -> DB: saveTorrentAndDownload()
+note right: DB save BEFORE qBit add\n(prevents orphan torrents)
+Workflow -> QBit: AddMagnet()
 
-Workflow -> QBit: AddMagnet(magnet_uri)
-QBit --> Workflow: OK
+Workflow -> DB: album_events.Create(COMPLETE)
+Workflow -> Bus: Publish(COMPLETE)
+Bus ->> Server: event notification
+Server ->> Client: MonitorAlbumResponse
+note right #lightgreen: Final result
 
-== 9. Persist & Schedule ==
+Workflow -> DB: workflow_runs.SetCompleted()
+Workflow -> Registry: Remove(albumID, quality)
 
-Workflow ->> Client: StatusUpdate(SAVING)
-note right #lightblue: "Saving to database..."
+== 4. Client Disconnect (Fire-and-Forget) ==
 
-Workflow -> DB: torrents.Create / downloads.Create
-Workflow -> River: Insert(PollDownloadArgs)
-
-== 10. Return Result ==
-
-Workflow ->> Client: StatusUpdate(COMPLETE)
-note right #lightblue: "Download started"
-
-Workflow ->> Client: MonitorAlbumResponse
-note right #lightgreen: Final result with:\nalbum, artist, release, download
-
-Server -> Server: Close stream
+note over Client, Workflow #lightyellow
+Client can disconnect at ANY point during the workflow.
+The workflow goroutine continues independently.
+Another client can subscribe to the same workflow later.
+end note
 
 @enduml
diff --git a/docs/architecture/flows/monitor-album-stream-manual.puml b/docs/architecture/flows/monitor-album-stream-manual.puml
index 7480595..4fcf944 100644
--- a/docs/architecture/flows/monitor-album-stream-manual.puml
+++ b/docs/architecture/flows/monitor-album-stream-manual.puml
@@ -113,6 +113,21 @@ note right #lightblue: "Download started"
 Workflow ->> Client: MonitorAlbumResponse
 note right #lightgreen: Final result
 
+== Cancel Cleanup (Disconnect or Cancel Message) ==
+
+note over Client, QBit #salmon
+**Manual Mode: Disconnect = Cancel**
+
+When client disconnects or sends CancelRequest:
+1. Workflow context is cancelled (stops further processing)
+2. If torrent was added to qBit: **DeleteTorrent(hash)** removes it + files
+3. If download record exists: marked as **cancelled** in DB
+4. workflow_run marked as **cancelled** in DB
+5. All events persisted to album_events for audit trail
+
+Cleanup uses a fresh context (not the cancelled one).
+end note
+
 == Decision Points Summary ==
 
 note over Client, QBit #lightyellow
diff --git a/docs/architecture/flows/monitor-album-stream-protocol.puml b/docs/architecture/flows/monitor-album-stream-protocol.puml
index 3079ab2..93bb3b9 100644
--- a/docs/architecture/flows/monitor-album-stream-protocol.puml
+++ b/docs/architecture/flows/monitor-album-stream-protocol.puml
@@ -1,11 +1,11 @@
 @startuml MonitorAlbumStream Protocol
 skinparam sequenceMessageAlign center
-title MonitorAlbumStream: Message Protocol
+title MonitorAlbumStream & SubscribeEvents: Message Protocol
 
 participant "Client" as C
 participant "Server" as S
 
-== Stream Initialization ==
+== MonitorAlbumStream (Bidirectional) ==
 
 C -> S: gRPC MonitorAlbumStream()
 note right: Opens bidirectional stream
@@ -50,7 +50,44 @@ note left #lightgreen
   - recoverable: bool
 end note
 
-== Monitor Steps (Status Updates) ==
+== SubscribeEvents (Server-Side Stream) ==
+
+C -> S: SubscribeEvents(SubscribeEventsRequest)
+note right #lightyellow
+**SubscribeEventsRequest:**
+- since_seq: int64 (0 = live only, >0 = replay from seq)
+
+**Response stream: AlbumEvent**
+- seq: int64 (monotonic sequence number)
+- workflow_run_id, album_id, quality
+- event_type: status | error | result
+- step: MonitorStep name
+- message: human-readable text
+- data_json: bytes (optional structured data)
+- timestamp_ms: int64
+
+**Global firehose**: receives events from ALL running workflows.
+Client-side filtering by album_id if needed.
+end note
+
+== Event Flow Architecture ==
+
+note over C, S #lightblue
+**Event Path (DB first, bus second):**
+
+1. Workflow step executes
+2. Event written to **album_events** table (synchronous, durable)
+3. Event published to **EventBus** (async, ephemeral notification)
+4. Bus fans out to subscribers:
+   - MonitorAlbumStream clients (per-topic)
+   - SubscribeEvents clients (global)
+5. Subscribers convert event to proto and stream.Send()
+
+**DB is source of truth. Bus is notification layer.**
+Events are never lost, even if no subscribers are connected.
+end note
+
+== Monitor Steps ==
 
 note over C, S #lightyellow
 **MonitorStep Enum:**
@@ -65,34 +102,34 @@ note over C, S #lightyellow
 9. COMPLETE - Workflow finished
 end note
 
-== Prompt Types (Manual Mode) ==
+== Mode Comparison ==
 
-note over C, S #orange
-**CONFIRM** (yes/no decision)
-- confirm_label, cancel_label
-- default_value: bool
+note over C, S #lightyellow
+**AUTOMATIC mode:**
+- Workflow runs as background goroutine
+- Client can disconnect, workflow continues (fire-and-forget)
+- Duplicate request for same album+quality subscribes to existing
+- Events delivered via EventBus bridge
 
-**SELECT_ONE** (pick one option)
-- options: [{id, label, description}]
-- default_id: pre-selected option
-
-**SELECT_MANY** (pick multiple options)
-- options: [{id, label, description}]
-- default_ids: pre-selected options
-- min_selections, max_selections
+**MANUAL mode:**
+- Workflow runs inline in stream handler
+- Interactive prompts at 4 decision points
+- Disconnect = cancel = full cleanup (qBit delete + DB cancel)
+- Events delivered directly via stream + persisted to DB
 end note
 
-== Data Payloads ==
+== Persistence ==
 
-note over C, S #lightblue
-**StreamAlbumInfo** (at FETCHING_METADATA)
-- artist, title, release_date, already_owned, owned_quality
+note over C, S #lightgreen
+**workflow_runs table:**
+- Tracks workflow lifecycle: running → completed | failed | cancelled
+- Unique constraint: one running workflow per album+quality
+- Used for deduplication, recovery, and audit
 
-**TorrentList** (at PARSING_RESULTS)
-- torrents: [{id, title, tracker, seeders, format, lossless}]
-
-**ReleaseInfo** (at SELECTING_RELEASE)
-- info_hash, format, bit_depth, sample_rate, seeders, tracker
+**album_events table:**
+- Audit log of all workflow events
+- seq BIGSERIAL for ordering and replay
+- Supports subscribe-before-query replay pattern
 end note
 
 == Timeout Behavior ==
@@ -102,9 +139,6 @@ When prompt times out (max: 300s):
 - Server uses **default decision** value
 - Workflow continues automatically
 - No error is raised
-
-Timeout is capped server-side at 300s.
-If timeout_seconds on prompt is 0 or exceeds max, 300s is used.
 end note
 
 @enduml