ios: show provider/model in subtitle

docker-compose.example.yml

@@ -12,6 +12,9 @@ services:
       OPENAI_API_KEY: ${OPENAI_API_KEY:-}
       ANTHROPIC_API_KEY: ${ANTHROPIC_API_KEY:-}
       XAI_API_KEY: ${XAI_API_KEY:-}
+      HERMES_AGENT_API_BASE_URL: ${HERMES_AGENT_API_BASE_URL:-http://127.0.0.1:8642/v1}
+      HERMES_AGENT_API_KEY: ${HERMES_AGENT_API_KEY:-}
+      HERMES_AGENT_MODEL: ${HERMES_AGENT_MODEL:-}
       EXA_API_KEY: ${EXA_API_KEY:-}
       CHAT_WEB_SEARCH_ENGINE: ${CHAT_WEB_SEARCH_ENGINE:-exa}
       SEARXNG_BASE_URL: ${SEARXNG_BASE_URL:-}

docs/api/rest.md

@@ -33,11 +33,29 @@ Chat upload limits:
   "providers": {
     "openai": { "models": ["gpt-4.1-mini"], "loadedAt": "2026-02-14T00:00:00.000Z", "error": null },
     "anthropic": { "models": ["claude-3-5-sonnet-latest"], "loadedAt": null, "error": null },
-    "xai": { "models": ["grok-3-mini"], "loadedAt": null, "error": null }
+    "xai": { "models": ["grok-3-mini"], "loadedAt": null, "error": null },
+    "hermes-agent": { "models": ["hermes-agent"], "loadedAt": null, "error": null }
   }
 }
 ```
 - OpenAI model lists are filtered to models that are expected to work with the backend's Responses API implementation.
+- `hermes-agent` is included only when `HERMES_AGENT_API_KEY` is configured. Set it to Hermes `API_SERVER_KEY`, or any non-empty value if that local server does not require auth. `HERMES_AGENT_API_BASE_URL` defaults to `http://127.0.0.1:8642/v1`; set `HERMES_AGENT_MODEL` only when you need an additional fallback/override model id.
+
+## Active Runs
+
+### `GET /v1/active-runs`
+- Response:
+```json
+{
+  "chats": ["chat-id-with-active-stream"],
+  "searches": ["search-id-with-active-stream"]
+}
+```
+
+Behavior notes:
+- Lists in-memory chat/search streams that are still running on this server process.
+- Clients should use this after app start or page refresh to restore per-row generating indicators.
+- The lists are not durable across server restarts.
 
 ## Chats
 
@@ -49,7 +67,7 @@ Chat upload limits:
 ```json
 {
   "title": "optional title",
-  "provider": "optional openai|anthropic|xai",
+  "provider": "optional openai|anthropic|xai|hermes-agent",
   "model": "optional model id",
   "messages": [
     {
@@ -136,7 +154,7 @@ Notes:
 ```json
 {
   "chatId": "optional-chat-id",
-  "provider": "openai|anthropic|xai",
+  "provider": "openai|anthropic|xai|hermes-agent",
   "model": "string",
   "messages": [
     {
@@ -190,11 +208,12 @@ Behavior notes:
 - Text files are forwarded as explicit text blocks rather than provider-managed file references. Large text attachments should already be truncated client-side before submission.
 - For `openai`, backend calls OpenAI's Responses API and enables internal tool use with an internal system instruction.
 - For `xai`, backend calls xAI's OpenAI-compatible Chat Completions API and enables internal tool use with the same internal system instruction.
+- For `hermes-agent`, backend calls the configured Hermes Agent OpenAI-compatible Chat Completions API without adding Sybil-managed tool definitions; Hermes Agent handles its own tools server-side.
 - For `openai`, image attachments are sent as Responses `input_image` items and text attachments are sent as `input_text` items.
-- For `xai`, image attachments are sent as Chat Completions content parts alongside text.
+- For `xai` and `hermes-agent`, image attachments are sent as Chat Completions content parts alongside text.
 - For `openai`, Responses calls that can enter the server-managed tool loop use `store: true` so reasoning and function-call items can be passed between tool rounds.
 - For `anthropic`, image attachments are sent as Messages API `image` blocks using base64 source data; text attachments are added as `text` blocks.
-- Available tool calls for chat: `web_search` and `fetch_url`. When `CHAT_CODEX_TOOL_ENABLED=true`, `codex_exec` is also available. When `CHAT_SHELL_TOOL_ENABLED=true`, `shell_exec` is also available.
+- Available Sybil-managed tool calls for `openai` and `xai`: `web_search` and `fetch_url`. When `CHAT_CODEX_TOOL_ENABLED=true`, `codex_exec` is also available. When `CHAT_SHELL_TOOL_ENABLED=true`, `shell_exec` is also available.
 - `web_search` returns ranked results with per-result summaries/snippets. Its backend engine is selected by `CHAT_WEB_SEARCH_ENGINE` (`exa` default, or `searxng` with `SEARXNG_BASE_URL` set). SearXNG mode requires the instance to allow `format=json`.
 - `fetch_url` fetches a URL and returns plaintext page content (HTML converted to text server-side).
 - `codex_exec` delegates coding, shell, repository inspection, and other complex software tasks to a persistent remote Codex CLI workspace over SSH. The server runs `codex exec --dangerously-bypass-approvals-and-sandbox --skip-git-repo-check <non-interactive wrapped prompt>` on the configured devbox inside `CHAT_CODEX_REMOTE_WORKDIR`, with SSH stdin closed.
@@ -260,6 +279,32 @@ Search run notes:
 - Persists answer text/citations + ranked results.
 - If both search and answer fail, endpoint returns an error.
 
+### `POST /v1/searches/:searchId/run/stream`
+- Body: same as `POST /v1/searches/:searchId/run`
+- Response: `text/event-stream`
+
+Events:
+- `search_results`: `{ "requestId": string|null, "results": SearchResultItem[] }`
+- `search_error`: `{ "error": string }`
+- `answer`: `{ "answerText": string|null, "answerRequestId": string|null, "answerCitations": SearchDetail["answerCitations"] }`
+- `answer_error`: `{ "error": string }`
+- terminal `done`: `{ "search": SearchDetail }`
+- terminal `error`: `{ "message": string }`
+
+Behavior notes:
+- The stream is owned by the backend after it starts. If the original HTTP client disconnects, the backend keeps running and persists the final search state.
+- While a search stream is active, `GET /v1/active-runs` includes the `searchId`.
+- If a stream is already active for the same `searchId`, this endpoint attaches to the existing stream instead of starting a second run.
+
+### `POST /v1/searches/:searchId/run/stream/attach`
+- Body: none
+- Response: `text/event-stream` with the same event names as `POST /v1/searches/:searchId/run/stream`
+- Not found: `404 { "message": "active search stream not found" }`
+
+Behavior notes:
+- Replays buffered events for the active in-memory stream, then emits new events until `done` or `error`.
+- Intended for clients that discovered a pending search via `GET /v1/active-runs`, such as after browser refresh.
+
 ## Type Shapes
 
 `ChatSummary`
@@ -269,9 +314,9 @@ Search run notes:
   "title": null,
   "createdAt": "...",
   "updatedAt": "...",
-  "initiatedProvider": "openai|anthropic|xai|null",
+  "initiatedProvider": "openai|anthropic|xai|hermes-agent|null",
   "initiatedModel": "string|null",
-  "lastUsedProvider": "openai|anthropic|xai|null",
+  "lastUsedProvider": "openai|anthropic|xai|hermes-agent|null",
   "lastUsedModel": "string|null"
 }
 ```
@@ -317,9 +362,9 @@ Search run notes:
   "title": null,
   "createdAt": "...",
   "updatedAt": "...",
-  "initiatedProvider": "openai|anthropic|xai|null",
+  "initiatedProvider": "openai|anthropic|xai|hermes-agent|null",
   "initiatedModel": "string|null",
-  "lastUsedProvider": "openai|anthropic|xai|null",
+  "lastUsedProvider": "openai|anthropic|xai|hermes-agent|null",
   "lastUsedModel": "string|null",
   "messages": [Message]
 }

docs/api/streaming-chat.md

@@ -4,6 +4,7 @@ This document defines the server-sent events (SSE) contract for chat completions
 
 Endpoint:
 - `POST /v1/chat-completions/stream`
+- `POST /v1/chats/:chatId/stream/attach`
 
 Transport:
 - HTTP response uses `Content-Type: text/event-stream; charset=utf-8`
@@ -20,7 +21,7 @@ Authentication:
 {
   "chatId": "optional-chat-id",
   "persist": true,
-  "provider": "openai|anthropic|xai",
+  "provider": "openai|anthropic|xai|hermes-agent",
   "model": "string",
   "messages": [
     {
@@ -61,6 +62,23 @@ Notes:
 - For persisted streams, backend stores only new non-assistant input history rows to avoid duplicates.
 - Attachments are optional and are persisted under `message.metadata.attachments` on stored user messages when `persist` is `true`.
 
+Persisted chat streams with a `chatId` are backend-owned active runs:
+- Once started, the backend keeps the stream running even if the HTTP client disconnects or refreshes.
+- While running, `GET /v1/active-runs` includes the `chatId`.
+- Starting a second persisted stream for the same active `chatId` returns `409`.
+- Clients can reattach with `POST /v1/chats/:chatId/stream/attach`.
+
+## Attach Endpoint
+
+`POST /v1/chats/:chatId/stream/attach`
+- Body: none.
+- Response uses the same `text/event-stream` transport and event names as `POST /v1/chat-completions/stream`.
+- Replays buffered events for the active in-memory stream, then emits new events until `done` or `error`.
+- Returns `404 { "message": "active chat stream not found" }` if no stream is currently active for that chat.
+- Authentication is the same as all other API endpoints.
+
+This endpoint is intended for clients that restored an active `chatId` from `GET /v1/active-runs`, especially after browser refresh. Replayed `delta` events may include text that was originally emitted before the client attached.
+
 ## Event Stream Contract
 
 Event order:
@@ -134,8 +152,9 @@ For `persist: false` streams, `chatId` and `callId` are `null`.
 
 - `openai`: backend uses OpenAI's Responses API and may execute internal function tool calls (`web_search`, `fetch_url`, optional `codex_exec`, and optional `shell_exec`) before producing final text.
 - `xai`: backend uses xAI's OpenAI-compatible Chat Completions API and may execute the same internal tool calls before producing final text.
+- `hermes-agent`: backend uses the configured Hermes Agent OpenAI-compatible Chat Completions API. Sybil does not add its own tool definitions for this provider; Hermes Agent handles its own tools server-side. Custom Hermes stream events are normalized away unless they produce text deltas in this SSE contract.
 - `openai`: image attachments are sent as Responses `input_image` items; text attachments are sent as `input_text` items.
-- `xai`: image attachments are sent as Chat Completions content parts; text attachments are inlined as text parts.
+- `xai` and `hermes-agent`: image attachments are sent as Chat Completions content parts; text attachments are inlined as text parts.
 - `openai`: Responses calls that can enter the server-managed tool loop use `store: true` so reasoning and function-call items can be passed between tool rounds.
 - `anthropic`: streamed via event stream; emits `delta` from `content_block_delta` with `text_delta`. Image attachments are sent as base64 `image` blocks and text attachments are appended as `text` blocks.
 - `web_search` uses `CHAT_WEB_SEARCH_ENGINE` (`exa` default, or `searxng` with `SEARXNG_BASE_URL` set). SearXNG mode requires the instance to allow `format=json`. This only affects chat-mode tool calls, not search-mode endpoints.

ios/AGENTS.md

@@ -51,3 +51,4 @@ Instructions for work under `/Users/buzzert/src/sybil-2/ios`.
   - OpenAI: `gpt-4.1-mini`
   - Anthropic: `claude-3-5-sonnet-latest`
   - xAI: `grok-3-mini`
+  - Hermes Agent: `hermes-agent`

ios/Apps/Sybil/Info.plist

@@ -0,0 +1,17 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<!DOCTYPE plist PUBLIC "-//Apple//DTD PLIST 1.0//EN" "http://www.apple.com/DTDs/PropertyList-1.0.dtd">
+<plist version="1.0">
+<dict>
+	<key>UIApplicationShortcutItems</key>
+	<array>
+		<dict>
+			<key>UIApplicationShortcutItemType</key>
+			<string>net.buzzert.sybil2.quick-question</string>
+			<key>UIApplicationShortcutItemTitle</key>
+			<string>Quick question</string>
+            <key>UIApplicationShortcutItemIconSymbolName</key>
+            <string>sparkles</string>
+		</dict>
+	</array>
+</dict>
+</plist>

ios/Apps/Sybil/Sources/SybilApp.swift

@@ -5,6 +5,8 @@ import UIKit
 @main
 struct SybilApp: App
 {
+    @UIApplicationDelegateAdaptor(SybilAppDelegate.self) private var appDelegate
+
     var body: some Scene {
         WindowGroup {
             SplitView()
@@ -14,3 +16,79 @@ struct SybilApp: App
         }
     }
 }
+
+@MainActor
+final class SybilAppDelegate: NSObject, UIApplicationDelegate {
+    func application(
+        _ application: UIApplication,
+        didFinishLaunchingWithOptions launchOptions: [UIApplication.LaunchOptionsKey: Any]? = nil
+    ) -> Bool {
+        SybilHomeScreenQuickActionHandler.configureQuickActions()
+        return true
+    }
+
+    func application(
+        _ application: UIApplication,
+        configurationForConnecting connectingSceneSession: UISceneSession,
+        options: UIScene.ConnectionOptions
+    ) -> UISceneConfiguration {
+        let configuration = UISceneConfiguration(
+            name: "Default Configuration",
+            sessionRole: connectingSceneSession.role
+        )
+        configuration.delegateClass = SybilSceneDelegate.self
+        return configuration
+    }
+
+    func application(
+        _ application: UIApplication,
+        performActionFor shortcutItem: UIApplicationShortcutItem,
+        completionHandler: @escaping (Bool) -> Void
+    ) {
+        completionHandler(SybilHomeScreenQuickActionHandler.handle(shortcutItem))
+    }
+}
+
+@MainActor
+final class SybilSceneDelegate: NSObject, UIWindowSceneDelegate {
+    func scene(
+        _ scene: UIScene,
+        willConnectTo session: UISceneSession,
+        options connectionOptions: UIScene.ConnectionOptions
+    ) {
+        if let shortcutItem = connectionOptions.shortcutItem {
+            _ = SybilHomeScreenQuickActionHandler.handle(shortcutItem)
+        }
+    }
+
+    func windowScene(
+        _ windowScene: UIWindowScene,
+        performActionFor shortcutItem: UIApplicationShortcutItem,
+        completionHandler: @escaping (Bool) -> Void
+    ) {
+        completionHandler(SybilHomeScreenQuickActionHandler.handle(shortcutItem))
+    }
+
+    func sceneWillResignActive(_ scene: UIScene) {
+        SybilHomeScreenQuickActionHandler.configureQuickActions()
+    }
+}
+
+@MainActor
+private enum SybilHomeScreenQuickActionHandler {
+    static func configureQuickActions() {
+        // The quick question action is static in Info.plist so it is available before first launch.
+        UIApplication.shared.shortcutItems = []
+    }
+
+    static func handle(_ shortcutItem: UIApplicationShortcutItem) -> Bool {
+        guard shortcutItem.type == SybilHomeScreenQuickAction.quickQuestionType else {
+            return false
+        }
+
+        Task { @MainActor in
+            SybilQuickActionRouter.shared.requestQuickQuestionPresentation()
+        }
+        return true
+    }
+}

ios/Apps/Sybil/project.yml

@@ -22,9 +22,10 @@ targets:
         SUPPORTS_MAC_DESIGNED_FOR_IPHONE_IPAD: NO
         TARGETED_DEVICE_FAMILY: "1,2,6"
         GENERATE_INFOPLIST_FILE: YES
+        INFOPLIST_FILE: Apps/Sybil/Info.plist
         ASSETCATALOG_COMPILER_APPICON_NAME: AppIcon
-        MARKETING_VERSION: 1.5
-        CURRENT_PROJECT_VERSION: 6
+        MARKETING_VERSION: 1.8
+        CURRENT_PROJECT_VERSION: 9
         INFOPLIST_KEY_CFBundleDisplayName: Sybil
         INFOPLIST_KEY_ITSAppUsesNonExemptEncryption: NO
         INFOPLIST_KEY_UIApplicationSupportsIndirectInputEvents: YES

ios/Packages/Sybil/Sources/Sybil/SplitView.swift

@@ -2,10 +2,14 @@ import SwiftUI
 
 public struct SplitView: View {
     @State private var viewModel = SybilViewModel()
+    @ObservedObject private var quickActionRouter = SybilQuickActionRouter.shared
     @Environment(\.horizontalSizeClass) private var horizontalSizeClass
     @Environment(\.scenePhase) private var scenePhase
     @State private var shouldRefreshOnForeground = false
     @State private var composerFocusRequest = 0
+    @State private var quickQuestionFocusRequest = 0
+    @State private var hasPendingQuickQuestionPresentation = false
+    @State private var isQuickQuestionPresented = false
     @State private var columnVisibility: NavigationSplitViewVisibility = .automatic
 
     private var keyboardActions: SybilKeyboardActions? {
@@ -74,8 +78,28 @@ public struct SplitView: View {
         .font(.sybil(.body))
         .preferredColorScheme(.dark)
         .focusedSceneValue(\.sybilKeyboardActions, keyboardActions)
+        .sheet(isPresented: $isQuickQuestionPresented, onDismiss: handleQuickQuestionDismissed) {
+            SybilQuickQuestionView(
+                viewModel: viewModel,
+                focusRequest: quickQuestionFocusRequest
+            )
+            .presentationDragIndicator(.visible)
+        }
         .task {
             await viewModel.bootstrap()
+            presentPendingQuickQuestionIfPossible()
+        }
+        .onReceive(quickActionRouter.$quickQuestionPresentationRequest) { request in
+            guard request > 0 else {
+                return
+            }
+            queueQuickQuestionPresentation()
+        }
+        .onChange(of: viewModel.isCheckingSession) { _, _ in
+            presentPendingQuickQuestionIfPossible()
+        }
+        .onChange(of: viewModel.isAuthenticated) { _, _ in
+            presentPendingQuickQuestionIfPossible()
         }
         .onChange(of: scenePhase) { _, nextPhase in
             switch nextPhase {
@@ -112,6 +136,28 @@ public struct SplitView: View {
             columnVisibility = .all
         }
     }
+
+    private func queueQuickQuestionPresentation() {
+        hasPendingQuickQuestionPresentation = true
+        presentPendingQuickQuestionIfPossible()
+    }
+
+    private func presentPendingQuickQuestionIfPossible() {
+        guard hasPendingQuickQuestionPresentation,
+              !viewModel.isCheckingSession,
+              viewModel.isAuthenticated
+        else {
+            return
+        }
+
+        hasPendingQuickQuestionPresentation = false
+        quickQuestionFocusRequest += 1
+        isQuickQuestionPresented = true
+    }
+
+    private func handleQuickQuestionDismissed() {
+        viewModel.cancelQuickQuestion()
+    }
 }
 
 public struct SybilCommands: Commands {

ios/Packages/Sybil/Sources/Sybil/SybilAPIClient.swift

@@ -49,11 +49,16 @@ actor SybilAPIClient: SybilAPIClienting {
         return response.chats
     }
 
-    func createChat(title: String? = nil) async throws -> ChatSummary {
+    func createChat(
+        title: String? = nil,
+        provider: Provider? = nil,
+        model: String? = nil,
+        messages: [CompletionRequestMessage]? = nil
+    ) async throws -> ChatSummary {
         let response = try await request(
             "/v1/chats",
             method: "POST",
-            body: AnyEncodable(ChatCreateBody(title: title)),
+            body: AnyEncodable(ChatCreateBody(title: title, provider: provider, model: model, messages: messages)),
             responseType: ChatCreateResponse.self
         )
         return response.chat
@@ -116,6 +121,10 @@ actor SybilAPIClient: SybilAPIClienting {
         try await request("/v1/models", method: "GET", responseType: ModelCatalogResponse.self)
     }
 
+    func getActiveRuns() async throws -> ActiveRunsResponse {
+        try await request("/v1/active-runs", method: "GET", responseType: ActiveRunsResponse.self)
+    }
+
     func runCompletionStream(
         body: CompletionStreamRequest,
         onEvent: @escaping @Sendable (CompletionStreamEvent) async -> Void
@@ -133,43 +142,35 @@ actor SybilAPIClient: SybilAPIClienting {
         )
 
         try await stream(request: request) { eventName, dataText in
-            switch eventName {
-            case "meta":
-                let payload: CompletionStreamMeta = try Self.decodeEvent(dataText, as: CompletionStreamMeta.self, eventName: eventName)
-                await onEvent(.meta(payload))
-            case "tool_call":
-                let payload: CompletionStreamToolCall = try Self.decodeEvent(dataText, as: CompletionStreamToolCall.self, eventName: eventName)
-                await onEvent(.toolCall(payload))
-            case "delta":
-                let payload: CompletionStreamDelta = try Self.decodeEvent(dataText, as: CompletionStreamDelta.self, eventName: eventName)
-                await onEvent(.delta(payload))
-            case "done":
-                do {
-                    let payload: CompletionStreamDone = try Self.decodeEvent(dataText, as: CompletionStreamDone.self, eventName: eventName)
-                    await onEvent(.done(payload))
-                } catch {
-                    if let recovered = Self.decodeLastJSONLine(dataText, as: CompletionStreamDone.self) {
-                        SybilLog.warning(
-                            SybilLog.network,
-                            "Recovered chat stream done payload from concatenated SSE data"
-                        )
-                        await onEvent(.done(recovered))
-                    } else {
-                        throw error
-                    }
-                }
-            case "error":
-                let payload: StreamErrorPayload = try Self.decodeEvent(dataText, as: StreamErrorPayload.self, eventName: eventName)
-                await onEvent(.error(payload))
-            default:
-                SybilLog.warning(SybilLog.network, "Ignoring unknown chat stream event '\(eventName)'")
-                await onEvent(.ignored)
-            }
+            try await Self.handleCompletionStreamEvent(eventName: eventName, dataText: dataText, onEvent: onEvent)
         }
 
         SybilLog.info(SybilLog.network, "Chat stream completed")
     }
 
+    func attachCompletionStream(
+        chatID: String,
+        onEvent: @escaping @Sendable (CompletionStreamEvent) async -> Void
+    ) async throws {
+        let request = try makeRequest(
+            path: "/v1/chats/\(chatID)/stream/attach",
+            method: "POST",
+            body: nil,
+            acceptsSSE: true
+        )
+
+        SybilLog.info(
+            SybilLog.network,
+            "Attaching chat stream POST \(request.url?.absoluteString ?? "<unknown>")"
+        )
+
+        try await stream(request: request) { eventName, dataText in
+            try await Self.handleCompletionStreamEvent(eventName: eventName, dataText: dataText, onEvent: onEvent)
+        }
+
+        SybilLog.info(SybilLog.network, "Attached chat stream completed")
+    }
+
     func runSearchStream(
         searchID: String,
         body: SearchRunRequest,
@@ -188,34 +189,35 @@ actor SybilAPIClient: SybilAPIClienting {
         )
 
         try await stream(request: request) { eventName, dataText in
-            switch eventName {
-            case "search_results":
-                let payload: SearchResultsPayload = try Self.decodeEvent(dataText, as: SearchResultsPayload.self, eventName: eventName)
-                await onEvent(.searchResults(payload))
-            case "search_error":
-                let payload: SearchErrorPayload = try Self.decodeEvent(dataText, as: SearchErrorPayload.self, eventName: eventName)
-                await onEvent(.searchError(payload))
-            case "answer":
-                let payload: SearchAnswerPayload = try Self.decodeEvent(dataText, as: SearchAnswerPayload.self, eventName: eventName)
-                await onEvent(.answer(payload))
-            case "answer_error":
-                let payload: SearchErrorPayload = try Self.decodeEvent(dataText, as: SearchErrorPayload.self, eventName: eventName)
-                await onEvent(.answerError(payload))
-            case "done":
-                let payload: SearchDonePayload = try Self.decodeEvent(dataText, as: SearchDonePayload.self, eventName: eventName)
-                await onEvent(.done(payload))
-            case "error":
-                let payload: StreamErrorPayload = try Self.decodeEvent(dataText, as: StreamErrorPayload.self, eventName: eventName)
-                await onEvent(.error(payload))
-            default:
-                SybilLog.warning(SybilLog.network, "Ignoring unknown search stream event '\(eventName)'")
-                await onEvent(.ignored)
-            }
+            try await Self.handleSearchStreamEvent(eventName: eventName, dataText: dataText, onEvent: onEvent)
         }
 
         SybilLog.info(SybilLog.network, "Search stream completed")
     }
 
+    func attachSearchStream(
+        searchID: String,
+        onEvent: @escaping @Sendable (SearchStreamEvent) async -> Void
+    ) async throws {
+        let request = try makeRequest(
+            path: "/v1/searches/\(searchID)/run/stream/attach",
+            method: "POST",
+            body: nil,
+            acceptsSSE: true
+        )
+
+        SybilLog.info(
+            SybilLog.network,
+            "Attaching search stream POST \(request.url?.absoluteString ?? "<unknown>")"
+        )
+
+        try await stream(request: request) { eventName, dataText in
+            try await Self.handleSearchStreamEvent(eventName: eventName, dataText: dataText, onEvent: onEvent)
+        }
+
+        SybilLog.info(SybilLog.network, "Attached search stream completed")
+    }
+
     private func request<Response: Decodable>(
         _ path: String,
         method: String,
@@ -498,6 +500,75 @@ actor SybilAPIClient: SybilAPIClienting {
         return try? Self.decodeJSON(type, from: data)
     }
 
+    private static func handleCompletionStreamEvent(
+        eventName: String,
+        dataText: String,
+        onEvent: @escaping @Sendable (CompletionStreamEvent) async -> Void
+    ) async throws {
+        switch eventName {
+        case "meta":
+            let payload: CompletionStreamMeta = try Self.decodeEvent(dataText, as: CompletionStreamMeta.self, eventName: eventName)
+            await onEvent(.meta(payload))
+        case "tool_call":
+            let payload: CompletionStreamToolCall = try Self.decodeEvent(dataText, as: CompletionStreamToolCall.self, eventName: eventName)
+            await onEvent(.toolCall(payload))
+        case "delta":
+            let payload: CompletionStreamDelta = try Self.decodeEvent(dataText, as: CompletionStreamDelta.self, eventName: eventName)
+            await onEvent(.delta(payload))
+        case "done":
+            do {
+                let payload: CompletionStreamDone = try Self.decodeEvent(dataText, as: CompletionStreamDone.self, eventName: eventName)
+                await onEvent(.done(payload))
+            } catch {
+                if let recovered = Self.decodeLastJSONLine(dataText, as: CompletionStreamDone.self) {
+                    SybilLog.warning(
+                        SybilLog.network,
+                        "Recovered chat stream done payload from concatenated SSE data"
+                    )
+                    await onEvent(.done(recovered))
+                } else {
+                    throw error
+                }
+            }
+        case "error":
+            let payload: StreamErrorPayload = try Self.decodeEvent(dataText, as: StreamErrorPayload.self, eventName: eventName)
+            await onEvent(.error(payload))
+        default:
+            SybilLog.warning(SybilLog.network, "Ignoring unknown chat stream event '\(eventName)'")
+            await onEvent(.ignored)
+        }
+    }
+
+    private static func handleSearchStreamEvent(
+        eventName: String,
+        dataText: String,
+        onEvent: @escaping @Sendable (SearchStreamEvent) async -> Void
+    ) async throws {
+        switch eventName {
+        case "search_results":
+            let payload: SearchResultsPayload = try Self.decodeEvent(dataText, as: SearchResultsPayload.self, eventName: eventName)
+            await onEvent(.searchResults(payload))
+        case "search_error":
+            let payload: SearchErrorPayload = try Self.decodeEvent(dataText, as: SearchErrorPayload.self, eventName: eventName)
+            await onEvent(.searchError(payload))
+        case "answer":
+            let payload: SearchAnswerPayload = try Self.decodeEvent(dataText, as: SearchAnswerPayload.self, eventName: eventName)
+            await onEvent(.answer(payload))
+        case "answer_error":
+            let payload: SearchErrorPayload = try Self.decodeEvent(dataText, as: SearchErrorPayload.self, eventName: eventName)
+            await onEvent(.answerError(payload))
+        case "done":
+            let payload: SearchDonePayload = try Self.decodeEvent(dataText, as: SearchDonePayload.self, eventName: eventName)
+            await onEvent(.done(payload))
+        case "error":
+            let payload: StreamErrorPayload = try Self.decodeEvent(dataText, as: StreamErrorPayload.self, eventName: eventName)
+            await onEvent(.error(payload))
+        default:
+            SybilLog.warning(SybilLog.network, "Ignoring unknown search stream event '\(eventName)'")
+            await onEvent(.ignored)
+        }
+    }
+
     private static func flushSSEEvent(
         eventName: inout String,
         dataLines: inout [String]
@@ -551,6 +622,7 @@ actor SybilAPIClient: SybilAPIClienting {
 
 struct CompletionStreamRequest: Codable, Sendable {
     var chatId: String?
+    var persist: Bool? = nil
     var provider: Provider
     var model: String
     var messages: [CompletionRequestMessage]
@@ -558,6 +630,9 @@ struct CompletionStreamRequest: Codable, Sendable {
 
 private struct ChatCreateBody: Encodable {
     var title: String?
+    var provider: Provider?
+    var model: String?
+    var messages: [CompletionRequestMessage]?
 }
 
 private struct SearchCreateBody: Encodable {

ios/Packages/Sybil/Sources/Sybil/SybilAPIClienting.swift

@@ -3,7 +3,12 @@ import Foundation
 protocol SybilAPIClienting: Sendable {
     func verifySession() async throws -> AuthSession
     func listChats() async throws -> [ChatSummary]
-    func createChat(title: String?) async throws -> ChatSummary
+    func createChat(
+        title: String?,
+        provider: Provider?,
+        model: String?,
+        messages: [CompletionRequestMessage]?
+    ) async throws -> ChatSummary
     func getChat(chatID: String) async throws -> ChatDetail
     func deleteChat(chatID: String) async throws
     func suggestChatTitle(chatID: String, content: String) async throws -> ChatSummary
@@ -13,13 +18,28 @@ protocol SybilAPIClienting: Sendable {
     func createChatFromSearch(searchID: String, title: String?) async throws -> ChatSummary
     func deleteSearch(searchID: String) async throws
     func listModels() async throws -> ModelCatalogResponse
+    func getActiveRuns() async throws -> ActiveRunsResponse
     func runCompletionStream(
         body: CompletionStreamRequest,
         onEvent: @escaping @Sendable (CompletionStreamEvent) async -> Void
     ) async throws
+    func attachCompletionStream(
+        chatID: String,
+        onEvent: @escaping @Sendable (CompletionStreamEvent) async -> Void
+    ) async throws
     func runSearchStream(
         searchID: String,
         body: SearchRunRequest,
         onEvent: @escaping @Sendable (SearchStreamEvent) async -> Void
     ) async throws
+    func attachSearchStream(
+        searchID: String,
+        onEvent: @escaping @Sendable (SearchStreamEvent) async -> Void
+    ) async throws
+}
+
+extension SybilAPIClienting {
+    func createChat(title: String?) async throws -> ChatSummary {
+        try await createChat(title: title, provider: nil, model: nil, messages: nil)
+    }
 }

ios/Packages/Sybil/Sources/Sybil/SybilChatTranscriptView.swift

@@ -17,18 +17,6 @@ struct SybilChatTranscriptView: View {
     var body: some View {
         ScrollView {
             LazyVStack(alignment: .leading, spacing: 26) {
-                if isSending && !hasPendingAssistant {
-                    HStack(spacing: 8) {
-                        ProgressView()
-                            .controlSize(.small)
-                            .tint(SybilTheme.textMuted)
-                        Text("Assistant is typing…")
-                            .font(.sybil(.footnote))
-                            .foregroundStyle(SybilTheme.textMuted)
-                    }
-                    .scaleEffect(x: 1, y: -1)
-                }
-
                 ForEach(messages.reversed()) { message in
                     MessageBubble(message: message, isSending: isSending)
                         .frame(maxWidth: .infinity)

ios/Packages/Sybil/Sources/Sybil/SybilModels.swift

@@ -4,12 +4,14 @@ public enum Provider: String, Codable, CaseIterable, Hashable, Sendable {
     case openai
     case anthropic
     case xai
+    case hermesAgent = "hermes-agent"
 
     public var displayName: String {
         switch self {
         case .openai: return "OpenAI"
         case .anthropic: return "Anthropic"
         case .xai: return "xAI"
+        case .hermesAgent: return "Hermes Agent"
         }
     }
 }
@@ -354,6 +356,16 @@ public struct SearchDetail: Codable, Identifiable, Hashable, Sendable {
     public var results: [SearchResultItem]
 }
 
+public struct ActiveRunsResponse: Codable, Hashable, Sendable {
+    public var chats: [String]
+    public var searches: [String]
+
+    public init(chats: [String] = [], searches: [String] = []) {
+        self.chats = chats
+        self.searches = searches
+    }
+}
+
 public struct SearchRunRequest: Codable, Sendable {
     public var query: String?
     public var title: String?
@@ -394,8 +406,8 @@ public struct CompletionRequestMessage: Codable, Sendable {
 }
 
 public struct CompletionStreamMeta: Codable, Sendable {
-    public var chatId: String
-    public var callId: String
+    public var chatId: String?
+    public var callId: String?
     public var provider: Provider
     public var model: String
 }

ios/Packages/Sybil/Sources/Sybil/SybilPhoneShellView.swift

@@ -22,75 +22,60 @@ enum PhoneRoute: Hashable {
 
 struct SybilPhoneShellView: View {
     @Bindable var viewModel: SybilViewModel
-    @State private var path: [PhoneRoute] = []
+    @State private var route: PhoneRoute = .draftChat
     @Environment(\.scenePhase) private var scenePhase
     @State private var shouldRefreshOnForeground = false
     @State private var composerFocusRequest = 0
     @State private var phoneStackWidth: CGFloat = BackSwipeMetrics.referenceWidth
-    @State private var backSwipeOffset: CGFloat = 0
-    @State private var backSwipeCompletionOffset: CGFloat = 0
-    @State private var backSwipeIsActive = false
-    @State private var backSwipeIsCompleting = false
-    @State private var backSwipeHasLatched = false
+    @State private var isSidebarOverlayPresented = false
+    @State private var sidebarSwipeOffset: CGFloat = 0
+    @State private var sidebarSwipeIsActive = false
+    @State private var sidebarSwipeIsCompleting = false
+    @State private var sidebarSwipeHasLatched = false
+    @State private var sidebarHighlightSelection: SidebarSelection?
+    @State private var sidebarHighlightClearTask: Task<Void, Never>?
+    @State private var openingSelectionRequestID: UUID?
 
-    private var canRecognizeBackSwipe: Bool {
-        !path.isEmpty && !backSwipeIsCompleting
+    private var canRecognizeSidebarSwipe: Bool {
+        !isSidebarOverlayPresented && !sidebarSwipeIsCompleting
     }
 
-    private var backSwipeVisualOffset: CGFloat {
-        backSwipeOffset + backSwipeCompletionOffset
+    private var sidebarOverlayProgress: CGFloat {
+        if isSidebarOverlayPresented {
+            return 1
+        }
+
+        return SidebarOverlaySwipeMetrics.progress(
+            for: sidebarSwipeOffset,
+            width: phoneStackWidth
+        )
+    }
+
+    private var shouldRenderSidebarOverlay: Bool {
+        isSidebarOverlayPresented ||
+            sidebarSwipeIsActive ||
+            sidebarSwipeIsCompleting ||
+            sidebarOverlayProgress > 0.001
+    }
+
+    private var currentRouteSelection: SidebarSelection? {
+        switch route {
+        case let .chat(chatID):
+            return .chat(chatID)
+        case let .search(searchID):
+            return .search(searchID)
+        case .draftChat, .draftSearch, .settings:
+            return nil
+        }
+    }
+
+    private var highlightedSidebarSelection: SidebarSelection? {
+        sidebarHighlightSelection ?? currentRouteSelection
     }
 
     var body: some View {
         GeometryReader { proxy in
-            ZStack(alignment: .topLeading) {
-                SybilPhoneSidebarRoot(viewModel: viewModel, path: $path)
-                    .safeAreaInset(edge: .top, spacing: 0) {
-                        phoneRootTopBar
-                    }
-                    .zIndex(0)
-
-                if let route = path.last {
-                    SybilPhoneDestinationView(
-                        viewModel: viewModel,
-                        composerFocusRequest: $composerFocusRequest,
-                        route: route,
-                        onRequestBack: requestBack,
-                        onRequestNewChat: startNewChatFromDestination
-                    )
-                    .background(SybilTheme.background)
-                    .offset(x: backSwipeVisualOffset)
-                    .shadow(
-                        color: backSwipeVisualOffset > 0 ? Color.black.opacity(0.34) : Color.clear,
-                        radius: backSwipeVisualOffset > 0 ? 18 : 0,
-                        x: -8,
-                        y: 0
-                    )
-                    .transition(.move(edge: .trailing))
-                    .zIndex(1)
-                    .background {
-                        WorkspaceSwipePanInstaller(
-                            direction: .right,
-                            isEnabled: canRecognizeBackSwipe,
-                            onBegan: { width in
-                                beginBackSwipe(containerWidth: width)
-                            },
-                            onChanged: { translationX, width in
-                                updateBackSwipe(with: translationX, containerWidth: width)
-                            },
-                            onEnded: { translationX, width, velocityX, didFinish in
-                                finishBackSwipe(
-                                    translationX: translationX,
-                                    containerWidth: width,
-                                    velocityX: velocityX,
-                                    didFinish: didFinish
-                                )
-                            }
-                        )
-                        .frame(maxWidth: .infinity, maxHeight: .infinity)
-                    }
-                }
-            }
+            phoneStack(width: proxy.size.width)
             .frame(maxWidth: .infinity, maxHeight: .infinity, alignment: .topLeading)
             .onAppear {
                 updatePhoneStackWidth(proxy.size.width)
@@ -100,13 +85,8 @@ struct SybilPhoneShellView: View {
             }
         }
         .tint(SybilTheme.primary)
-        .animation(.easeOut(duration: 0.22), value: path.last)
-        .onChange(of: path) { _, nextPath in
-            guard nextPath.isEmpty else {
-                return
-            }
-            resetBackSwipe(animated: false)
-        }
+        .animation(.easeOut(duration: 0.22), value: route)
+        .animation(.easeOut(duration: 0.18), value: isSidebarOverlayPresented)
         .onChange(of: scenePhase) { _, nextPhase in
             switch nextPhase {
             case .background:
@@ -120,8 +100,8 @@ struct SybilPhoneShellView: View {
                 shouldRefreshOnForeground = false
                 Task {
                     await viewModel.refreshAfterAppBecameActive(
-                        refreshCollections: path.isEmpty,
-                        refreshSelection: !path.isEmpty && viewModel.hasRefreshableSelection
+                        refreshCollections: isSidebarOverlayPresented,
+                        refreshSelection: !isSidebarOverlayPresented && viewModel.hasRefreshableSelection
                     )
                 }
             case .inactive:
@@ -133,16 +113,117 @@ struct SybilPhoneShellView: View {
         }
     }
 
-    private var phoneRootTopBar: some View {
-        HStack {
+    private func phoneStack(width: CGFloat) -> some View {
+        ZStack(alignment: .topLeading) {
+            phoneWorkspaceLayer
+                .zIndex(0)
+
+            phoneSidebarOverlayLayer(width: width)
+                .zIndex(1)
+        }
+    }
+
+    private var phoneWorkspaceLayer: some View {
+        SybilPhoneDestinationView(
+            viewModel: viewModel,
+            composerFocusRequest: $composerFocusRequest,
+            route: route,
+            onRequestBack: { _ in showSidebarOverlay() },
+            onRequestNewChat: sidebarWorkspaceNewChatAction,
+            onShowSidebar: showSidebarOverlay
+        )
+        .background(SybilTheme.background)
+        .blur(radius: SidebarOverlaySwipeMetrics.workspaceBlurRadius(for: sidebarOverlayProgress))
+        .opacity(SidebarOverlaySwipeMetrics.workspaceOpacity(for: sidebarOverlayProgress))
+        .allowsHitTesting(!shouldRenderSidebarOverlay)
+        .background {
+            sidebarSwipeInstaller
+        }
+    }
+
+    private func phoneSidebarOverlayLayer(width: CGFloat) -> some View {
+        VStack(spacing: 0) {
+            phoneOverlayTopBar
+
+            SybilPhoneSidebarRoot(
+                viewModel: viewModel,
+                highlightedSelection: highlightedSidebarSelection,
+                onSelect: openSidebarSelection,
+                onRoute: showRouteAndClearSidebarHighlight
+            )
+        }
+        .opacity(sidebarOverlayProgress)
+        .blur(radius: SidebarOverlaySwipeMetrics.overlayBlurRadius(for: sidebarOverlayProgress))
+        .offset(x: SidebarOverlaySwipeMetrics.overlayOffset(for: sidebarOverlayProgress, width: width))
+        .allowsHitTesting(isSidebarOverlayPresented)
+        .accessibilityHidden(!isSidebarOverlayPresented)
+    }
+
+    private var sidebarSwipeInstaller: some View {
+        WorkspaceSwipePanInstaller(
+            direction: .right,
+            isEnabled: canRecognizeSidebarSwipe,
+            onBegan: { width in
+                beginSidebarSwipe(containerWidth: width)
+            },
+            onChanged: { translationX, width in
+                updateSidebarSwipe(with: translationX, containerWidth: width)
+            },
+            onEnded: { translationX, width, velocityX, didFinish in
+                finishSidebarSwipe(
+                    translationX: translationX,
+                    containerWidth: width,
+                    velocityX: velocityX,
+                    didFinish: didFinish
+                )
+            }
+        )
+        .frame(maxWidth: .infinity, maxHeight: .infinity)
+    }
+
+    private var sidebarWorkspaceNewChatAction: (() -> Void)? {
+        guard !isSidebarOverlayPresented else {
+            return nil
+        }
+
+        return {
+            startNewChatFromDestination()
+        }
+    }
+
+    private var phoneOverlayTopBar: some View {
+        HStack(spacing: 12) {
             SybilWordmark(size: 21)
             Spacer()
+
+            Button {
+                hideSidebarOverlay()
+            } label: {
+                Image(systemName: "chevron.right.2")
+                    .font(.system(size: 21, weight: .bold))
+                    .foregroundStyle(SybilTheme.text)
+                    .frame(width: 54, height: 54)
+                    .background(
+                        Circle()
+                            .fill(.ultraThinMaterial)
+                            .overlay(
+                                Circle()
+                                    .fill(SybilTheme.surface.opacity(0.76))
+                            )
+                    )
+                    .overlay(
+                        Circle()
+                            .stroke(SybilTheme.border.opacity(0.64), lineWidth: 1)
+                    )
+            }
+            .buttonStyle(.plain)
+            .accessibilityLabel("Hide conversations")
         }
         .padding(.horizontal, 16)
         .padding(.top, 10)
         .padding(.bottom, 12)
         .background {
-            SybilTheme.panelGradient
+            SybilPhoneOverlayBlurBand(edge: .top)
                 .ignoresSafeArea(edges: .top)
         }
     }
@@ -151,62 +232,98 @@ struct SybilPhoneShellView: View {
         phoneStackWidth = max(width, 1)
     }
 
-    private func requestBack(animateNavigation: Bool = true) {
-        guard !path.isEmpty, !backSwipeIsCompleting else {
-            return
-        }
-
-        if animateNavigation {
-            Task {
-                await completeBackSwipe(containerWidth: phoneStackWidth, releaseVelocityX: 0)
-            }
-        } else {
-            popRoute(disablesAnimations: true)
-            resetBackSwipe(animated: false)
-        }
-    }
-
     private func startNewChatFromDestination() {
         viewModel.startNewChat()
         composerFocusRequest += 1
-        replaceTopRoute(with: .draftChat)
+        showRoute(.draftChat)
     }
 
-    private func replaceTopRoute(with route: PhoneRoute) {
-        if path.isEmpty {
+    private func showRoute(_ nextRoute: PhoneRoute) {
+        let update = {
+            route = nextRoute
+        }
+
+        if isSidebarOverlayPresented {
             withAnimation(.easeOut(duration: 0.22)) {
-                path = [route]
+                update()
+                isSidebarOverlayPresented = false
             }
         } else {
-            path[path.index(before: path.endIndex)] = route
+            update()
         }
+
+        resetSidebarSwipe(animated: false)
     }
 
-    private func popRoute(disablesAnimations: Bool) {
-        let pop = {
-            guard !path.isEmpty else {
+    private func showRouteAndClearSidebarHighlight(_ nextRoute: PhoneRoute) {
+        showRoute(nextRoute)
+        clearSidebarHighlight()
+    }
+
+    private func showSidebarOverlay() {
+        withAnimation(.easeOut(duration: 0.18)) {
+            isSidebarOverlayPresented = true
+        }
+        resetSidebarSwipe(animated: false)
+    }
+
+    private func hideSidebarOverlay() {
+        withAnimation(.easeOut(duration: 0.18)) {
+            isSidebarOverlayPresented = false
+        }
+        resetSidebarSwipe(animated: false)
+    }
+
+    private func openSidebarSelection(_ selection: SidebarSelection) {
+        if openingSelectionRequestID != nil, sidebarHighlightSelection == selection {
+            return
+        }
+
+        let requestID = UUID()
+        openingSelectionRequestID = requestID
+        setSidebarHighlight(selection)
+
+        Task {
+            await viewModel.selectForNavigation(selection)
+            guard openingSelectionRequestID == requestID else {
                 return
             }
-            _ = path.removeLast()
-        }
 
-        if disablesAnimations {
-            var transaction = Transaction()
-            transaction.disablesAnimations = true
-            withTransaction(transaction) {
-                pop()
-            }
-        } else {
-            withAnimation(.easeOut(duration: 0.22)) {
-                pop()
-            }
+            showRoute(PhoneRoute.from(selection: selection))
+            openingSelectionRequestID = nil
+            clearSidebarHighlight(selection, after: .milliseconds(260))
         }
     }
 
-    private func beginBackSwipe(containerWidth: CGFloat) {
+    private func setSidebarHighlight(_ selection: SidebarSelection) {
+        sidebarHighlightClearTask?.cancel()
+        sidebarHighlightSelection = selection
+    }
+
+    private func clearSidebarHighlight(_ selection: SidebarSelection, after delay: Duration) {
+        sidebarHighlightClearTask?.cancel()
+        sidebarHighlightClearTask = Task { @MainActor in
+            try? await Task.sleep(for: delay)
+            guard !Task.isCancelled,
+                  sidebarHighlightSelection == selection,
+                  openingSelectionRequestID == nil else {
+                return
+            }
+            sidebarHighlightSelection = nil
+        }
+    }
+
+    private func clearSidebarHighlight() {
+        sidebarHighlightClearTask?.cancel()
+        openingSelectionRequestID = nil
+        sidebarHighlightSelection = nil
+    }
+
+    private func beginSidebarSwipe(containerWidth: CGFloat) {
         let update = {
-            backSwipeIsActive = true
-            backSwipeHasLatched = false
+            phoneStackWidth = max(containerWidth, 1)
+            sidebarSwipeIsActive = true
+            sidebarSwipeHasLatched = false
         }
 
         var transaction = Transaction()
@@ -214,97 +331,79 @@ struct SybilPhoneShellView: View {
         withTransaction(transaction, update)
     }
 
-    private func updateBackSwipe(with rawTranslation: CGFloat, containerWidth: CGFloat) {
-        let nextOffset = BackSwipeMetrics.clampedOffset(for: rawTranslation, width: containerWidth)
-        let nextLatched = BackSwipeMetrics.isLatched(
+    private func updateSidebarSwipe(with rawTranslation: CGFloat, containerWidth: CGFloat) {
+        let nextOffset = SidebarOverlaySwipeMetrics.clampedOffset(for: rawTranslation, width: containerWidth)
+        let nextLatched = SidebarOverlaySwipeMetrics.isLatched(
             offset: nextOffset,
             width: containerWidth,
-            isCurrentlyLatched: backSwipeHasLatched
+            isCurrentlyLatched: sidebarSwipeHasLatched
         )
 
         var transaction = Transaction()
         transaction.disablesAnimations = true
         withTransaction(transaction) {
-            backSwipeOffset = nextOffset
-            backSwipeHasLatched = nextLatched
+            phoneStackWidth = max(containerWidth, 1)
+            sidebarSwipeOffset = nextOffset
+            sidebarSwipeHasLatched = nextLatched
         }
     }
 
-    private func finishBackSwipe(
+    private func finishSidebarSwipe(
         translationX: CGFloat,
         containerWidth: CGFloat,
         velocityX: CGFloat,
         didFinish: Bool
     ) {
-        guard backSwipeIsActive else {
-            resetBackSwipe(animated: false)
+        guard sidebarSwipeIsActive else {
+            resetSidebarSwipe(animated: false)
             return
         }
 
-        let finalOffset = BackSwipeMetrics.clampedOffset(for: translationX, width: containerWidth)
-        let finalLatched = BackSwipeMetrics.isLatched(
+        let finalOffset = SidebarOverlaySwipeMetrics.clampedOffset(for: translationX, width: containerWidth)
+        let finalLatched = SidebarOverlaySwipeMetrics.isLatched(
             offset: finalOffset,
             width: containerWidth,
-            isCurrentlyLatched: backSwipeHasLatched
+            isCurrentlyLatched: sidebarSwipeHasLatched
         )
-        updateBackSwipe(with: translationX, containerWidth: containerWidth)
+        updateSidebarSwipe(with: translationX, containerWidth: containerWidth)
 
-        if didFinish && BackSwipeMetrics.shouldComplete(
+        if didFinish && SidebarOverlaySwipeMetrics.shouldComplete(
             offset: finalOffset,
             velocityX: velocityX,
             width: containerWidth,
             isLatched: finalLatched
         ) {
-            Task {
-                await completeBackSwipe(containerWidth: containerWidth, releaseVelocityX: velocityX)
-            }
+            completeSidebarSwipe()
             return
         }
 
-        resetBackSwipe(animated: true, velocityX: velocityX)
+        resetSidebarSwipe(animated: true, velocityX: velocityX)
     }
 
-    @MainActor
-    private func completeBackSwipe(containerWidth: CGFloat, releaseVelocityX: CGFloat) async {
-        guard !path.isEmpty else {
-            resetBackSwipe(animated: false)
-            return
-        }
-        guard !backSwipeIsCompleting else {
+    private func completeSidebarSwipe() {
+        guard !sidebarSwipeIsCompleting else {
             return
         }
 
-        backSwipeIsCompleting = true
-        let targetOffset = BackSwipeMetrics.completionTargetOffset(for: containerWidth)
-
-        withAnimation(
-            BackSwipeMetrics.springAnimation(
-                currentOffset: backSwipeOffset,
-                targetOffset: targetOffset,
-                velocityX: releaseVelocityX
-            )
-        ) {
-            backSwipeCompletionOffset = targetOffset - backSwipeOffset
+        sidebarSwipeIsCompleting = true
+        withAnimation(.easeOut(duration: 0.18)) {
+            isSidebarOverlayPresented = true
         }
-
-        try? await Task.sleep(for: .milliseconds(BackSwipeMetrics.completionAnimationDelayMs))
-        popRoute(disablesAnimations: true)
-        resetBackSwipe(animated: false)
+        resetSidebarSwipe(animated: false)
     }
 
-    private func resetBackSwipe(animated: Bool, velocityX: CGFloat = 0) {
-        let currentOffset = backSwipeOffset + backSwipeCompletionOffset
+    private func resetSidebarSwipe(animated: Bool, velocityX: CGFloat = 0) {
+        let currentOffset = sidebarSwipeOffset
         let reset = {
-            backSwipeOffset = 0
-            backSwipeCompletionOffset = 0
-            backSwipeIsActive = false
-            backSwipeIsCompleting = false
-            backSwipeHasLatched = false
+            sidebarSwipeOffset = 0
+            sidebarSwipeIsActive = false
+            sidebarSwipeIsCompleting = false
+            sidebarSwipeHasLatched = false
         }
 
         if animated {
             withAnimation(
-                BackSwipeMetrics.springAnimation(
+                SidebarOverlaySwipeMetrics.springAnimation(
                     currentOffset: currentOffset,
                     targetOffset: 0,
                     velocityX: velocityX
@@ -318,31 +417,79 @@ struct SybilPhoneShellView: View {
     }
 }
 
-private struct SybilPhoneSidebarRoot: View {
-    @Bindable var viewModel: SybilViewModel
-    @Binding var path: [PhoneRoute]
-    @State private var openingSelection: SidebarSelection?
-    @State private var openingRequestID: UUID?
+private enum SidebarOverlaySwipeMetrics {
+    static func clampedOffset(for rawTranslation: CGFloat, width: CGFloat) -> CGFloat {
+        BackSwipeMetrics.clampedOffset(for: rawTranslation, width: width)
+    }
 
-    private var highlightedSelection: SidebarSelection? {
-        if let openingSelection {
-            return openingSelection
-        }
+    static func progress(for offset: CGFloat, width: CGFloat) -> CGFloat {
+        BackSwipeMetrics.progress(for: offset, width: width)
+    }
 
-        guard let route = path.last else {
-            return nil
-        }
+    static func isLatched(offset: CGFloat, width: CGFloat, isCurrentlyLatched: Bool = false) -> Bool {
+        BackSwipeMetrics.isLatched(offset: offset, width: width, isCurrentlyLatched: isCurrentlyLatched)
+    }
 
-        switch route {
-        case let .chat(chatID):
-            return .chat(chatID)
-        case let .search(searchID):
-            return .search(searchID)
-        case .draftChat, .draftSearch, .settings:
-            return nil
+    static func shouldComplete(offset: CGFloat, velocityX: CGFloat, width: CGFloat, isLatched: Bool) -> Bool {
+        BackSwipeMetrics.shouldComplete(offset: offset, velocityX: velocityX, width: width, isLatched: isLatched)
+    }
+
+    static func springAnimation(currentOffset: CGFloat, targetOffset: CGFloat, velocityX: CGFloat) -> Animation {
+        BackSwipeMetrics.springAnimation(currentOffset: currentOffset, targetOffset: targetOffset, velocityX: velocityX)
+    }
+
+    static func overlayOffset(for progress: CGFloat, width: CGFloat) -> CGFloat {
+        -(1 - min(max(progress, 0), 1)) * min(max(width * 0.18, 44), 76)
+    }
+
+    static func overlayBlurRadius(for progress: CGFloat) -> CGFloat {
+        (1 - min(max(progress, 0), 1)) * 18
+    }
+
+    static func workspaceBlurRadius(for progress: CGFloat) -> CGFloat {
+        min(max(progress, 0), 1) * 14
+    }
+
+    static func workspaceOpacity(for progress: CGFloat) -> CGFloat {
+        1 - (min(max(progress, 0), 1) * 0.22)
+    }
+}
+
+private struct SybilPhoneOverlayBlurBand: View {
+    var edge: VerticalEdge
+
+    var body: some View {
+        ZStack {
+            Rectangle()
+                .fill(.ultraThinMaterial)
+                .opacity(0.34)
+
+            Rectangle()
+                .fill(
+                    LinearGradient(
+                        colors: gradientColors,
+                        startPoint: edge == .top ? .top : .bottom,
+                        endPoint: edge == .top ? .bottom : .top
+                    )
+                )
         }
     }
 
+    private var gradientColors: [Color] {
+        [
+            Color.black.opacity(0.94),
+            SybilTheme.background.opacity(0.78),
+            Color.black.opacity(0)
+        ]
+    }
+}
+
+private struct SybilPhoneSidebarRoot: View {
+    @Bindable var viewModel: SybilViewModel
+    var highlightedSelection: SidebarSelection?
+    var onSelect: (SidebarSelection) -> Void
+    var onRoute: (PhoneRoute) -> Void
+
     var body: some View {
         VStack(spacing: 0) {
             if let errorMessage = viewModel.errorMessage {
@@ -357,64 +504,15 @@ private struct SybilPhoneSidebarRoot: View {
                     .overlay(SybilTheme.border)
             }
 
-            if viewModel.isLoadingCollections && viewModel.sidebarItems.isEmpty {
-                VStack(alignment: .leading, spacing: 8) {
-                    ProgressView()
-                        .tint(SybilTheme.primary)
-                    Text("Loading conversations…")
-                        .font(.sybil(.footnote))
-                        .foregroundStyle(SybilTheme.textMuted)
+            SybilSidebarItemList(
+                viewModel: viewModel,
+                isSelected: { item in
+                    highlightedSelection == item.selection
+                },
+                onSelect: { item in
+                    onSelect(item.selection)
                 }
-                .frame(maxWidth: .infinity, maxHeight: .infinity, alignment: .topLeading)
-                .padding(16)
-            } else if viewModel.sidebarItems.isEmpty {
-                VStack(spacing: 10) {
-                    Image(systemName: "message.badge")
-                        .font(.system(size: 20, weight: .medium))
-                        .foregroundStyle(SybilTheme.textMuted)
-                    Text("Start a chat or run your first search.")
-                        .font(.sybil(.footnote))
-                        .multilineTextAlignment(.center)
-                        .foregroundStyle(SybilTheme.textMuted)
-                }
-                .frame(maxWidth: .infinity, maxHeight: .infinity)
-                .padding(16)
-            } else {
-                ScrollView {
-                    LazyVStack(alignment: .leading, spacing: 0) {
-                        ForEach(viewModel.sidebarItems) { item in
-                            Button {
-                                open(item.selection)
-                            } label: {
-                                VStack(spacing: 0.0) {
-                                    SybilPhoneSidebarRow(item: item)
-                                    Divider()
-                                }
-                            }
-                            .buttonStyle(
-                                SybilPhoneSidebarRowButtonStyle(
-                                    isHighlighted: highlightedSelection == item.selection
-                                )
-                            )
-                            .contextMenu {
-                                Button(role: .destructive) {
-                                    Task {
-                                        await viewModel.deleteItem(item.selection)
-                                    }
-                                } label: {
-                                    Label("Delete", systemImage: "trash")
-                                }
-                            }
-                        }
-                    }
-                }
-                .refreshable {
-                    await viewModel.refreshVisibleContent(
-                        refreshCollections: true,
-                        refreshSelection: false
-                    )
-                }
-            }
+            )
         }
         .background(SybilTheme.panelGradient)
         .safeAreaInset(edge: .bottom, spacing: 0) {
@@ -429,22 +527,20 @@ private struct SybilPhoneSidebarRoot: View {
 
             HStack(spacing: 12) {
                 toolbarIconButton(systemImage: "gearshape", accessibilityLabel: "Settings") {
-                    clearOpeningSelection()
-                    showRoute(.settings)
+                    viewModel.openSettings()
+                    onRoute(.settings)
                 }
 
                 Spacer()
 
                 toolbarIconButton(systemImage: "magnifyingglass", accessibilityLabel: "New search") {
-                    clearOpeningSelection()
                     viewModel.startNewSearch()
-                    showRoute(.draftSearch)
+                    onRoute(.draftSearch)
                 }
 
                 toolbarIconButton(systemImage: "plus", accessibilityLabel: "New chat", isPrimary: true) {
-                    clearOpeningSelection()
                     viewModel.startNewChat()
-                    showRoute(.draftChat)
+                    onRoute(.draftChat)
                 }
             }
             .padding(.horizontal, 18)
@@ -480,114 +576,6 @@ private struct SybilPhoneSidebarRoot: View {
         .buttonStyle(.plain)
         .accessibilityLabel(accessibilityLabel)
     }
-
-    private func clearOpeningSelection() {
-        openingRequestID = nil
-        openingSelection = nil
-    }
-
-    private func showRoute(_ route: PhoneRoute) {
-        withAnimation(.easeOut(duration: 0.22)) {
-            path = [route]
-        }
-    }
-
-    private func open(_ selection: SidebarSelection) {
-        guard openingSelection != selection else {
-            return
-        }
-
-        let requestID = UUID()
-        openingRequestID = requestID
-        openingSelection = selection
-        Task {
-            await viewModel.selectForNavigation(selection)
-            guard openingRequestID == requestID else {
-                return
-            }
-            showRoute(PhoneRoute.from(selection: selection))
-            openingRequestID = nil
-            openingSelection = nil
-        }
-    }
-}
-
-private struct SybilPhoneSidebarRowIsActiveKey: EnvironmentKey {
-    static let defaultValue = false
-}
-
-private extension EnvironmentValues {
-    var sybilPhoneSidebarRowIsActive: Bool {
-        get { self[SybilPhoneSidebarRowIsActiveKey.self] }
-        set { self[SybilPhoneSidebarRowIsActiveKey.self] = newValue }
-    }
-}
-
-private struct SybilPhoneSidebarRowButtonStyle: ButtonStyle {
-    var isHighlighted: Bool
-
-    func makeBody(configuration: Configuration) -> some View {
-        configuration.label
-            .environment(\.sybilPhoneSidebarRowIsActive, isHighlighted || configuration.isPressed)
-    }
-}
-
-private struct SybilPhoneSidebarRow: View {
-    @Environment(\.sybilPhoneSidebarRowIsActive) private var isHighlighted
-    var item: SidebarItem
-
-    var body: some View {
-        let leadingWidth = 22.0
-        
-        VStack(alignment: .leading, spacing: 8) {
-            HStack(spacing: 8) {
-                Image(systemName: item.kind == .chat ? "message" : "globe")
-                    .font(.system(size: 12, weight: .semibold))
-                    .foregroundStyle(isHighlighted ? SybilTheme.accent : SybilTheme.textMuted)
-                    .frame(width: leadingWidth, height: leadingWidth)
-                    .background(
-                        Rectangle()
-                            .fill(isHighlighted ? SybilTheme.accent.opacity(0.12) : SybilTheme.surface.opacity(0.72))
-                            
-                    )
-
-                Text(item.title)
-                    .font(.sybil(.subheadline, weight: .semibold))
-                    .lineLimit(1)
-            }
-
-            HStack(spacing: 8) {
-                Spacer()
-                    .frame(width: leadingWidth)
-
-                Text(item.updatedAt.sybilRelativeLabel)
-                    .font(.sybil(.caption2))
-                    .foregroundStyle(SybilTheme.textMuted)
-
-                if let initiated = item.initiatedLabel {
-                    Spacer(minLength: 0)
-                    Text(initiated)
-                        .font(.sybil(.caption2))
-                        .foregroundStyle(SybilTheme.textMuted.opacity(0.88))
-                        .lineLimit(1)
-                        .multilineTextAlignment(.trailing)
-                        .frame(maxWidth: .infinity, alignment: .trailing)
-                }
-            }
-        }
-        .foregroundStyle(SybilTheme.text)
-        .padding(18.0)
-        .frame(maxWidth: .infinity, alignment: .leading)
-        .background(
-            Rectangle()
-                .fill(
-                    isHighlighted
-                        ? SybilTheme.selectedRowGradient
-                        : LinearGradient(colors: [SybilTheme.surface.opacity(0.56), SybilTheme.surface.opacity(0.36)], startPoint: .topLeading, endPoint: .bottomTrailing)
-                )
-        )
-
-    }
 }
 
 private struct SybilPhoneDestinationView: View {
@@ -595,12 +583,15 @@ private struct SybilPhoneDestinationView: View {
     @Binding var composerFocusRequest: Int
     let route: PhoneRoute
     let onRequestBack: (_ animateNavigation: Bool) -> Void
-    let onRequestNewChat: () -> Void
+    let onRequestNewChat: (() -> Void)?
+    let onShowSidebar: () -> Void
 
     var body: some View {
         SybilWorkspaceView(
             viewModel: viewModel,
             composerFocusRequest: composerFocusRequest,
+            navigationLeadingControl: .showSidebar,
+            onShowSidebar: onShowSidebar,
             onRequestBack: onRequestBack,
             onRequestNewChat: onRequestNewChat
         )

ios/Packages/Sybil/Sources/Sybil/SybilQuickActionRouting.swift

@@ -0,0 +1,19 @@
+import Combine
+import Foundation
+
+public enum SybilHomeScreenQuickAction {
+    public static let quickQuestionType = "net.buzzert.sybil2.quick-question"
+}
+
+@MainActor
+public final class SybilQuickActionRouter: ObservableObject {
+    public static let shared = SybilQuickActionRouter()
+
+    @Published public private(set) var quickQuestionPresentationRequest = 0
+
+    private init() {}
+
+    public func requestQuickQuestionPresentation() {
+        quickQuestionPresentationRequest += 1
+    }
+}

ios/Packages/Sybil/Sources/Sybil/SybilQuickQuestionView.swift

@@ -0,0 +1,302 @@
+import MarkdownUI
+import Observation
+import SwiftUI
+
+struct SybilQuickQuestionView: View {
+    @Bindable var viewModel: SybilViewModel
+    var focusRequest: Int
+
+    @Environment(\.dismiss) private var dismiss
+    @FocusState private var promptFocused: Bool
+
+    private var hasAnswerContent: Bool {
+        !viewModel.quickQuestionMessages.isEmpty || viewModel.quickQuestionError != nil
+    }
+
+    var body: some View {
+        VStack(spacing: 0) {
+            VStack(alignment: .leading, spacing: 16) {
+                header
+
+                answerArea
+
+                composer
+            }
+            .padding(.horizontal, 16)
+            .padding(.top, 18)
+            .padding(.bottom, 12)
+            .frame(maxWidth: 640, maxHeight: .infinity, alignment: .top)
+        }
+        .frame(maxWidth: .infinity, maxHeight: .infinity, alignment: .top)
+        .background(SybilTheme.backgroundGradient)
+        .preferredColorScheme(.dark)
+        .task(id: focusRequest) {
+            try? await Task.sleep(for: .milliseconds(260))
+            guard !Task.isCancelled else {
+                return
+            }
+            promptFocused = true
+        }
+    }
+
+    private var header: some View {
+        HStack {
+            Image(systemName: "sparkles")
+                .font(.system(size: 21, weight: .semibold))
+                .foregroundStyle(SybilTheme.primary)
+            
+            Text("Quick question")
+                .font(.title3.weight(.semibold))
+                .foregroundStyle(SybilTheme.text)
+                .lineLimit(1)
+            
+        }
+        .frame(maxWidth: .infinity, alignment: .leading)
+    }
+
+    private var answerArea: some View {
+        ScrollView {
+            VStack(alignment: .leading, spacing: 12) {
+                if hasAnswerContent {
+                    ForEach(viewModel.quickQuestionMessages) { message in
+                        QuickQuestionMessageView(message: message, isSending: viewModel.isQuickQuestionSending)
+                    }
+
+                    if let error = viewModel.quickQuestionError {
+                        Text(error)
+                            .font(.caption)
+                            .foregroundStyle(SybilTheme.danger)
+                            .fixedSize(horizontal: false, vertical: true)
+                    }
+                }
+            }
+            .frame(maxWidth: .infinity, alignment: .topLeading)
+            .padding(14)
+        }
+        .scrollDismissesKeyboard(.interactively)
+        .frame(maxWidth: .infinity, maxHeight: .infinity, alignment: .topLeading)
+        .background(
+            RoundedRectangle(cornerRadius: 12)
+                .fill(Color.black.opacity(0.36))
+        )
+        .overlay(
+            RoundedRectangle(cornerRadius: 12)
+                .stroke(SybilTheme.border.opacity(0.55), lineWidth: 1)
+        )
+    }
+
+    private var composer: some View {
+        VStack(alignment: .leading, spacing: 10) {
+            HStack(alignment: .bottom, spacing: 10) {
+                TextField(
+                    "Ask anything...",
+                    text: Binding(
+                        get: { viewModel.quickQuestionPrompt },
+                        set: { viewModel.updateQuickQuestionPrompt($0) }
+                    ),
+                    axis: .vertical
+                )
+                .focused($promptFocused)
+                .font(.body)
+                .textInputAutocapitalization(.sentences)
+                .autocorrectionDisabled(false)
+                .lineLimit(1 ... 6)
+                .submitLabel(.send)
+                .onSubmit(submitQuestion)
+                .padding(.horizontal, 12)
+                .padding(.vertical, 10)
+                .background(
+                    RoundedRectangle(cornerRadius: 12)
+                        .fill(SybilTheme.composerGradient)
+                        .opacity(0.98)
+                )
+                .foregroundStyle(SybilTheme.text)
+
+                Button(action: submitQuestion) {
+                    Image(systemName: "arrow.up")
+                        .font(.body.weight(.semibold))
+                        .frame(width: 40, height: 40)
+                        .background(
+                            Circle()
+                                .fill(
+                                    viewModel.canSendQuickQuestion
+                                        ? AnyShapeStyle(SybilTheme.primaryGradient)
+                                        : AnyShapeStyle(SybilTheme.surfaceStrong.opacity(0.92))
+                                )
+                        )
+                        .foregroundStyle(viewModel.canSendQuickQuestion ? SybilTheme.text : SybilTheme.textMuted)
+                }
+                .buttonStyle(.plain)
+                .disabled(!viewModel.canSendQuickQuestion)
+                .accessibilityLabel("Ask quick question")
+            }
+
+            controlsRow
+        }
+    }
+
+    private var convertButton: some View {
+        Button {
+            Task {
+                let didConvert = await viewModel.convertQuickQuestionToChat()
+                if didConvert {
+                    dismiss()
+                }
+            }
+        } label: {
+            Label("Chat", systemImage: "bubble.left")
+                .font(.caption.weight(.medium))
+                .lineLimit(1)
+                .minimumScaleFactor(0.8)
+        }
+        .buttonStyle(.plain)
+        .foregroundStyle(viewModel.canConvertQuickQuestion ? SybilTheme.text : SybilTheme.textMuted)
+        .padding(.horizontal, 10)
+        .frame(maxWidth: .infinity, minHeight: 40)
+        .background(
+            RoundedRectangle(cornerRadius: 12)
+                .fill(SybilTheme.surfaceStrong.opacity(0.78))
+                .overlay(
+                    RoundedRectangle(cornerRadius: 12)
+                        .stroke(SybilTheme.border.opacity(0.78), lineWidth: 1)
+                )
+        )
+        .disabled(!viewModel.canConvertQuickQuestion)
+    }
+
+    private var controlsRow: some View {
+        HStack(alignment: .center, spacing: 10) {
+            providerMenu
+            modelMenu
+            convertButton
+        }
+    }
+
+    private var providerMenu: some View {
+        Menu {
+            ForEach(viewModel.providerOptions, id: \.self) { provider in
+                Button {
+                    viewModel.setQuickQuestionProvider(provider)
+                } label: {
+                    if viewModel.quickQuestionProvider == provider {
+                        Label(provider.displayName, systemImage: "checkmark")
+                    } else {
+                        Text(provider.displayName)
+                    }
+                }
+            }
+        } label: {
+            QuickQuestionPickerPill(title: viewModel.quickQuestionProvider.displayName)
+        }
+        .frame(maxWidth: .infinity)
+        .disabled(viewModel.isQuickQuestionSending || viewModel.isConvertingQuickQuestion)
+        .accessibilityLabel("Quick question provider")
+    }
+
+    private var modelMenu: some View {
+        Menu {
+            if viewModel.quickQuestionProviderModelOptions.isEmpty {
+                Text("No models")
+            } else {
+                ForEach(viewModel.quickQuestionProviderModelOptions, id: \.self) { model in
+                    Button {
+                        viewModel.setQuickQuestionModel(model)
+                    } label: {
+                        if viewModel.quickQuestionModel == model {
+                            Label(model, systemImage: "checkmark")
+                        } else {
+                            Text(model)
+                        }
+                    }
+                }
+            }
+        } label: {
+            QuickQuestionPickerPill(title: viewModel.quickQuestionModel.isEmpty ? "No model" : viewModel.quickQuestionModel)
+        }
+        .frame(maxWidth: .infinity)
+        .disabled(viewModel.isQuickQuestionSending || viewModel.isConvertingQuickQuestion)
+        .accessibilityLabel("Quick question model")
+    }
+
+    private func submitQuestion() {
+        guard viewModel.canSendQuickQuestion else {
+            return
+        }
+
+        promptFocused = false
+        _ = viewModel.sendQuickQuestion()
+    }
+}
+
+private struct QuickQuestionPickerPill: View {
+    var title: String
+
+    var body: some View {
+        HStack(spacing: 8) {
+            Text(title)
+                .font(.caption.weight(.medium))
+                .foregroundStyle(SybilTheme.text)
+                .lineLimit(1)
+                .minimumScaleFactor(0.8)
+
+            Image(systemName: "chevron.down")
+                .font(.caption.weight(.semibold))
+                .foregroundStyle(SybilTheme.textMuted)
+        }
+        .padding(.horizontal, 10)
+        .frame(maxWidth: .infinity, minHeight: 40)
+        .background(
+            RoundedRectangle(cornerRadius: 12)
+                .fill(SybilTheme.surfaceStrong.opacity(0.78))
+                .overlay(
+                    RoundedRectangle(cornerRadius: 12)
+                        .stroke(SybilTheme.border.opacity(0.78), lineWidth: 1)
+                )
+        )
+    }
+}
+
+private struct QuickQuestionMessageView: View {
+    var message: Message
+    var isSending: Bool
+
+    private var isPendingAssistant: Bool {
+        message.id.hasPrefix("temp-assistant-quick-") &&
+            isSending &&
+            message.content.trimmingCharacters(in: .whitespacesAndNewlines).isEmpty
+    }
+
+    var body: some View {
+        if let metadata = message.toolCallMetadata {
+            Text(toolCallSummary(for: metadata, fallbackContent: message.content))
+                .font(.caption)
+                .foregroundStyle(SybilTheme.textMuted)
+                .fixedSize(horizontal: false, vertical: true)
+        } else if isPendingAssistant {
+            HStack(spacing: 8) {
+                ProgressView()
+                    .controlSize(.small)
+                    .tint(SybilTheme.primary)
+                Text("Thinking...")
+                    .font(.caption)
+                    .foregroundStyle(SybilTheme.textMuted)
+            }
+        } else if !message.content.trimmingCharacters(in: .whitespacesAndNewlines).isEmpty {
+            Markdown(message.content)
+                .font(.body)
+                .tint(SybilTheme.primary)
+                .foregroundStyle(SybilTheme.text.opacity(0.96))
+                .textSelection(.enabled)
+        }
+    }
+
+    private func toolCallSummary(for metadata: ToolCallMetadata, fallbackContent: String) -> String {
+        if let summary = metadata.summary?.trimmingCharacters(in: .whitespacesAndNewlines), !summary.isEmpty {
+            return summary
+        }
+        if !fallbackContent.trimmingCharacters(in: .whitespacesAndNewlines).isEmpty {
+            return fallbackContent
+        }
+        return "Ran \(metadata.toolName ?? "tool")."
+    }
+}

ios/Packages/Sybil/Sources/Sybil/SybilSettingsStore.swift

@@ -11,6 +11,12 @@ final class SybilSettingsStore {
         static let preferredOpenAIModel = "sybil.ios.preferredOpenAIModel"
         static let preferredAnthropicModel = "sybil.ios.preferredAnthropicModel"
         static let preferredXAIModel = "sybil.ios.preferredXAIModel"
+        static let preferredHermesAgentModel = "sybil.ios.preferredHermesAgentModel"
+        static let quickQuestionPreferredProvider = "sybil.ios.quickQuestionPreferredProvider"
+        static let quickQuestionPreferredOpenAIModel = "sybil.ios.quickQuestionPreferredOpenAIModel"
+        static let quickQuestionPreferredAnthropicModel = "sybil.ios.quickQuestionPreferredAnthropicModel"
+        static let quickQuestionPreferredXAIModel = "sybil.ios.quickQuestionPreferredXAIModel"
+        static let quickQuestionPreferredHermesAgentModel = "sybil.ios.quickQuestionPreferredHermesAgentModel"
     }
 
     private let defaults: UserDefaults
@@ -19,6 +25,8 @@ final class SybilSettingsStore {
     var adminToken: String
     var preferredProvider: Provider
     var preferredModelByProvider: [Provider: String]
+    var quickQuestionPreferredProvider: Provider
+    var quickQuestionPreferredModelByProvider: [Provider: String]
 
     init(defaults: UserDefaults = .standard) {
         self.defaults = defaults
@@ -32,10 +40,21 @@ final class SybilSettingsStore {
         let provider = defaults.string(forKey: Keys.preferredProvider).flatMap(Provider.init(rawValue:)) ?? .openai
         self.preferredProvider = provider
 
-        self.preferredModelByProvider = [
+        let preferredModels: [Provider: String] = [
             .openai: defaults.string(forKey: Keys.preferredOpenAIModel) ?? "gpt-4.1-mini",
             .anthropic: defaults.string(forKey: Keys.preferredAnthropicModel) ?? "claude-3-5-sonnet-latest",
-            .xai: defaults.string(forKey: Keys.preferredXAIModel) ?? "grok-3-mini"
+            .xai: defaults.string(forKey: Keys.preferredXAIModel) ?? "grok-3-mini",
+            .hermesAgent: defaults.string(forKey: Keys.preferredHermesAgentModel) ?? "hermes-agent"
+        ]
+        self.preferredModelByProvider = preferredModels
+
+        self.quickQuestionPreferredProvider =
+            defaults.string(forKey: Keys.quickQuestionPreferredProvider).flatMap(Provider.init(rawValue:)) ?? provider
+        self.quickQuestionPreferredModelByProvider = [
+            .openai: defaults.string(forKey: Keys.quickQuestionPreferredOpenAIModel) ?? preferredModels[.openai] ?? "gpt-4.1-mini",
+            .anthropic: defaults.string(forKey: Keys.quickQuestionPreferredAnthropicModel) ?? preferredModels[.anthropic] ?? "claude-3-5-sonnet-latest",
+            .xai: defaults.string(forKey: Keys.quickQuestionPreferredXAIModel) ?? preferredModels[.xai] ?? "grok-3-mini",
+            .hermesAgent: defaults.string(forKey: Keys.quickQuestionPreferredHermesAgentModel) ?? preferredModels[.hermesAgent] ?? "hermes-agent"
         ]
     }
 
@@ -53,6 +72,13 @@ final class SybilSettingsStore {
         defaults.set(preferredModelByProvider[.openai], forKey: Keys.preferredOpenAIModel)
         defaults.set(preferredModelByProvider[.anthropic], forKey: Keys.preferredAnthropicModel)
         defaults.set(preferredModelByProvider[.xai], forKey: Keys.preferredXAIModel)
+        defaults.set(preferredModelByProvider[.hermesAgent], forKey: Keys.preferredHermesAgentModel)
+
+        defaults.set(quickQuestionPreferredProvider.rawValue, forKey: Keys.quickQuestionPreferredProvider)
+        defaults.set(quickQuestionPreferredModelByProvider[.openai], forKey: Keys.quickQuestionPreferredOpenAIModel)
+        defaults.set(quickQuestionPreferredModelByProvider[.anthropic], forKey: Keys.quickQuestionPreferredAnthropicModel)
+        defaults.set(quickQuestionPreferredModelByProvider[.xai], forKey: Keys.quickQuestionPreferredXAIModel)
+        defaults.set(quickQuestionPreferredModelByProvider[.hermesAgent], forKey: Keys.quickQuestionPreferredHermesAgentModel)
     }
 
     var trimmedTokenOrNil: String? {
@@ -68,7 +94,7 @@ final class SybilSettingsStore {
             raw.removeLast()
         }
 
-        guard var components = URLComponents(string: raw) else {
+        guard let components = URLComponents(string: raw) else {
             return nil
         }
 

ios/Packages/Sybil/Sources/Sybil/SybilSidebarView.swift

@@ -4,13 +4,6 @@ import SwiftUI
 struct SybilSidebarView: View {
     @Bindable var viewModel: SybilViewModel
 
-    private func iconName(for item: SidebarItem) -> String {
-        switch item.kind {
-        case .chat: return "message"
-        case .search: return "globe"
-        }
-    }
-
     private func isSelected(_ item: SidebarItem) -> Bool {
         viewModel.draftKind == nil && viewModel.selectedItem == item.selection
     }
@@ -57,105 +50,13 @@ struct SybilSidebarView: View {
                     .overlay(SybilTheme.border)
             }
 
-            if viewModel.isLoadingCollections && viewModel.sidebarItems.isEmpty {
-                VStack(alignment: .leading, spacing: 8) {
-                    ProgressView()
-                        .tint(SybilTheme.primary)
-                    Text("Loading conversations…")
-                        .font(.sybil(.footnote))
-                        .foregroundStyle(SybilTheme.textMuted)
+            SybilSidebarItemList(
+                viewModel: viewModel,
+                isSelected: isSelected,
+                onSelect: { item in
+                    viewModel.select(item.selection)
                 }
-                .frame(maxWidth: .infinity, maxHeight: .infinity, alignment: .topLeading)
-                .padding(16)
-            } else if viewModel.sidebarItems.isEmpty {
-                VStack(spacing: 10) {
-                    Image(systemName: "message.badge")
-                        .font(.system(size: 20, weight: .medium))
-                        .foregroundStyle(SybilTheme.textMuted)
-                    Text("Start a chat or run your first search.")
-                        .font(.sybil(.footnote))
-                        .multilineTextAlignment(.center)
-                        .foregroundStyle(SybilTheme.textMuted)
-                }
-                .frame(maxWidth: .infinity, maxHeight: .infinity)
-                .padding(16)
-            } else {
-                ScrollView {
-                    LazyVStack(alignment: .leading, spacing: 8) {
-                        ForEach(viewModel.sidebarItems) { item in
-                            Button {
-                                viewModel.select(item.selection)
-                            } label: {
-                                VStack(alignment: .leading, spacing: 6) {
-                                    HStack(spacing: 8) {
-                                        Image(systemName: iconName(for: item))
-                                            .font(.system(size: 12, weight: .semibold))
-                                            .foregroundStyle(isSelected(item) ? SybilTheme.accent : SybilTheme.textMuted)
-                                            .frame(width: 22, height: 22)
-                                            .background(
-                                                RoundedRectangle(cornerRadius: 7)
-                                                    .fill(isSelected(item) ? SybilTheme.accent.opacity(0.12) : SybilTheme.surface.opacity(0.72))
-                                                    .overlay(
-                                                        RoundedRectangle(cornerRadius: 7)
-                                                            .stroke(isSelected(item) ? SybilTheme.accent.opacity(0.36) : SybilTheme.border.opacity(0.72), lineWidth: 1)
-                                                    )
-                                            )
-
-                                        Text(item.title)
-                                            .font(.sybil(.subheadline, weight: .semibold))
-                                            .lineLimit(1)
-                                    }
-
-                                    HStack(spacing: 8) {
-                                        Text(item.updatedAt.sybilRelativeLabel)
-                                            .font(.sybil(.caption2))
-                                            .foregroundStyle(SybilTheme.textMuted)
-
-                                        if let initiated = item.initiatedLabel {
-                                            Spacer(minLength: 0)
-                                            Text(initiated)
-                                                .font(.sybil(.caption2))
-                                                .foregroundStyle(SybilTheme.textMuted.opacity(0.88))
-                                                .lineLimit(1)
-                                                .multilineTextAlignment(.trailing)
-                                                .frame(maxWidth: .infinity, alignment: .trailing)
-                                        }
-                                    }
-                                }
-                                .foregroundStyle(SybilTheme.text)
-                                .padding(.horizontal, 12)
-                                .padding(.vertical, 10)
-                                .frame(maxWidth: .infinity, alignment: .leading)
-                                .background(
-                                    RoundedRectangle(cornerRadius: 12)
-                                        .fill(isSelected(item) ? SybilTheme.selectedRowGradient : LinearGradient(colors: [SybilTheme.surface.opacity(0.56), SybilTheme.surface.opacity(0.36)], startPoint: .topLeading, endPoint: .bottomTrailing))
-                                )
-                                .overlay(
-                                    RoundedRectangle(cornerRadius: 12)
-                                        .stroke(isSelected(item) ? SybilTheme.primary.opacity(0.55) : SybilTheme.border.opacity(0.72), lineWidth: 1)
-                                )
-                            }
-                            .buttonStyle(.plain)
-                            .contextMenu {
-                                Button(role: .destructive) {
-                                    Task {
-                                        await viewModel.deleteItem(item.selection)
-                                    }
-                                } label: {
-                                    Label("Delete", systemImage: "trash")
-                                }
-                            }
-                        }
-                    }
-                    .padding(10)
-                }
-                .refreshable {
-                    await viewModel.refreshVisibleContent(
-                        refreshCollections: true,
-                        refreshSelection: false
-                    )
-                }
-            }
+            )
 
         }
         .background(SybilTheme.panelGradient)
@@ -205,3 +106,148 @@ struct SybilSidebarView: View {
         .buttonStyle(.plain)
     }
 }
+
+struct SybilSidebarItemList: View {
+    @Bindable var viewModel: SybilViewModel
+    var isSelected: (SidebarItem) -> Bool
+    var onSelect: (SidebarItem) -> Void
+
+    var body: some View {
+        if viewModel.isLoadingCollections && viewModel.sidebarItems.isEmpty {
+            VStack(alignment: .leading, spacing: 8) {
+                ProgressView()
+                    .tint(SybilTheme.primary)
+                Text("Loading conversations…")
+                    .font(.sybil(.footnote))
+                    .foregroundStyle(SybilTheme.textMuted)
+            }
+            .frame(maxWidth: .infinity, maxHeight: .infinity, alignment: .topLeading)
+            .padding(16)
+        } else if viewModel.sidebarItems.isEmpty {
+            VStack(spacing: 10) {
+                Image(systemName: "message.badge")
+                    .font(.system(size: 20, weight: .medium))
+                    .foregroundStyle(SybilTheme.textMuted)
+                Text("Start a chat or run your first search.")
+                    .font(.sybil(.footnote))
+                    .multilineTextAlignment(.center)
+                    .foregroundStyle(SybilTheme.textMuted)
+            }
+            .frame(maxWidth: .infinity, maxHeight: .infinity)
+            .padding(16)
+        } else {
+            ScrollView {
+                LazyVStack(alignment: .leading, spacing: 8) {
+                    ForEach(viewModel.sidebarItems) { item in
+                        Button {
+                            onSelect(item)
+                        } label: {
+                            SybilSidebarRow(item: item, isSelected: isSelected(item))
+                        }
+                        .buttonStyle(.plain)
+                        .contextMenu {
+                            Button(role: .destructive) {
+                                Task {
+                                    await viewModel.deleteItem(item.selection)
+                                }
+                            } label: {
+                                Label("Delete", systemImage: "trash")
+                            }
+                        }
+                    }
+                }
+                .padding(10)
+            }
+            .refreshable {
+                await viewModel.refreshSidebarCollectionsFromPullToRefresh()
+            }
+        }
+    }
+}
+
+struct SybilSidebarRow: View {
+    var item: SidebarItem
+    var isSelected: Bool
+
+    private var isHighlighted: Bool {
+        isSelected
+    }
+
+    private var iconName: String {
+        switch item.kind {
+        case .chat: return "message"
+        case .search: return "globe"
+        }
+    }
+
+    var body: some View {
+        VStack(alignment: .leading, spacing: 6) {
+            HStack(spacing: 8) {
+                Image(systemName: iconName)
+                    .font(.system(size: 12, weight: .semibold))
+                    .foregroundStyle(isHighlighted ? SybilTheme.accent : SybilTheme.textMuted)
+                    .frame(width: 22, height: 22)
+                    .background(
+                        RoundedRectangle(cornerRadius: 7)
+                            .fill(isHighlighted ? SybilTheme.accent.opacity(0.12) : SybilTheme.surface.opacity(0.72))
+                            .overlay(
+                                RoundedRectangle(cornerRadius: 7)
+                                    .stroke(isHighlighted ? SybilTheme.accent.opacity(0.36) : SybilTheme.border.opacity(0.72), lineWidth: 1)
+                            )
+                    )
+
+                Text(item.title)
+                    .font(.sybil(.subheadline, weight: .semibold))
+                    .lineLimit(1)
+                    .layoutPriority(1)
+
+                Spacer(minLength: 8)
+
+                if item.isRunning {
+                    SybilSidebarActivityIndicator()
+                }
+            }
+
+            HStack(spacing: 8) {
+                Text(item.updatedAt.sybilRelativeLabel)
+                    .font(.sybil(.caption2))
+                    .foregroundStyle(SybilTheme.textMuted)
+
+                if let initiated = item.initiatedLabel {
+                    Spacer(minLength: 0)
+                    Text(initiated)
+                        .font(.sybil(.caption2))
+                        .foregroundStyle(SybilTheme.textMuted.opacity(0.88))
+                        .lineLimit(1)
+                        .multilineTextAlignment(.trailing)
+                        .frame(maxWidth: .infinity, alignment: .trailing)
+                }
+            }
+        }
+        .foregroundStyle(SybilTheme.text)
+        .padding(.horizontal, 12)
+        .padding(.vertical, 10)
+        .frame(maxWidth: .infinity, alignment: .leading)
+        .background(
+            RoundedRectangle(cornerRadius: 12)
+                .fill(isHighlighted ? SybilTheme.selectedRowGradient : LinearGradient(colors: [SybilTheme.surface.opacity(0.56), SybilTheme.surface.opacity(0.36)], startPoint: .topLeading, endPoint: .bottomTrailing))
+        )
+        .overlay(
+            RoundedRectangle(cornerRadius: 12)
+                .stroke(isHighlighted ? SybilTheme.primary.opacity(0.55) : SybilTheme.border.opacity(0.72), lineWidth: 1)
+        )
+        .contentShape(RoundedRectangle(cornerRadius: 12))
+    }
+}
+
+struct SybilSidebarActivityIndicator: View {
+    var body: some View {
+        ProgressView()
+            .progressViewStyle(.circular)
+            .controlSize(.small)
+            .tint(SybilTheme.accent)
+            .scaleEffect(0.82)
+            .frame(width: 16, height: 16)
+            .accessibilityLabel("Generating")
+    }
+}

ios/Packages/Sybil/Sources/Sybil/SybilWorkspaceView.swift

@@ -50,7 +50,7 @@ struct SybilWorkspaceView: View {
     }
 
     private var showsCustomWorkspaceNavigation: Bool {
-        usesCustomWorkspaceNavigation && (!isSettingsSelected || navigationLeadingControl == .back)
+        usesCustomWorkspaceNavigation && (!isSettingsSelected || navigationLeadingControl != .hidden)
     }
 
     private var transcriptScrollContextID: String {
@@ -75,7 +75,7 @@ struct SybilWorkspaceView: View {
         guard onRequestNewChat != nil else {
             return false
         }
-        guard !viewModel.isSending, viewModel.draftKind == nil else {
+        guard !viewModel.isActiveSelectionSending, viewModel.draftKind == nil else {
             return false
         }
         guard case .chat = viewModel.selectedItem else {
@@ -155,7 +155,7 @@ struct SybilWorkspaceView: View {
             workspaceContentStack
 
             if showsCustomWorkspaceNavigation {
-                SybilWorkspaceCharacterBackdrop(isBusy: viewModel.isSending)
+                SybilWorkspaceCharacterBackdrop(isBusy: viewModel.isActiveSelectionSending)
                     .allowsHitTesting(false)
                 customWorkspaceNavigationBar
             }
@@ -232,13 +232,7 @@ struct SybilWorkspaceView: View {
         HStack(spacing: 14) {
             workspaceNavigationLeadingControl
 
-            Text(viewModel.selectedTitle)
-                .font(.sybil(size: 16, weight: .semibold))
-                .foregroundStyle(SybilTheme.text)
-                .lineLimit(1)
-                .minimumScaleFactor(0.78)
-                .frame(maxWidth: .infinity, alignment: .leading)
-                .multilineTextAlignment(.leading)
+            customWorkspaceNavigationTitle
 
             workspaceNavigationTrailingControl
         }
@@ -251,6 +245,32 @@ struct SybilWorkspaceView: View {
         }
     }
 
+    private var selectedProviderModelSubtitle: String {
+        let selectedModel = viewModel.model.trimmingCharacters(in: .whitespacesAndNewlines)
+        guard !selectedModel.isEmpty else {
+            return viewModel.provider.displayName
+        }
+        return "\(viewModel.provider.displayName) • \(selectedModel)"
+    }
+
+    private var customWorkspaceNavigationTitle: some View {
+        VStack(alignment: .leading, spacing: 2) {
+            Text(viewModel.selectedTitle)
+                .font(.sybil(size: 16, weight: .semibold))
+                .foregroundStyle(SybilTheme.text)
+                .lineLimit(1)
+                .minimumScaleFactor(0.78)
+
+            Text(selectedProviderModelSubtitle)
+                .font(.sybil(size: 10, weight: .medium))
+                .foregroundStyle(SybilTheme.textMuted)
+                .lineLimit(1)
+                .minimumScaleFactor(0.82)
+        }
+        .frame(maxWidth: .infinity, alignment: .leading)
+        .multilineTextAlignment(.leading)
+    }
+
     @ViewBuilder
     private var workspaceNavigationLeadingControl: some View {
         switch navigationLeadingControl {
@@ -495,7 +515,7 @@ struct SybilWorkspaceView: View {
 
         Divider()
 
-        ForEach(Provider.allCases, id: \.self) { candidate in
+        ForEach(viewModel.providerOptions, id: \.self) { candidate in
             Menu(candidate.displayName) {
                 let models = viewModel.modelOptions(for: candidate)
                 if models.isEmpty {
@@ -560,10 +580,10 @@ struct SybilWorkspaceView: View {
                                 Circle()
                                     .stroke(SybilTheme.border.opacity(0.82), lineWidth: 1)
                             )
-                            .foregroundStyle(viewModel.isSending ? SybilTheme.textMuted : SybilTheme.text)
+                            .foregroundStyle(viewModel.isActiveSelectionSending ? SybilTheme.textMuted : SybilTheme.text)
                     }
                     .buttonStyle(.plain)
-                    .disabled(viewModel.isSending)
+                    .disabled(viewModel.isActiveSelectionSending)
                     .accessibilityLabel("Attach file")
                 }
 
@@ -626,7 +646,7 @@ struct SybilWorkspaceView: View {
             }
         }
         .onDrop(of: [UTType.fileURL.identifier, UTType.image.identifier], isTargeted: $isComposerDropTargeted) { providers in
-            if viewModel.isSearchMode || viewModel.isSending {
+            if viewModel.isSearchMode || viewModel.isActiveSelectionSending {
                 return false
             }
 
@@ -703,9 +723,7 @@ struct SybilWorkspaceView: View {
         }
 
         #if !targetEnvironment(macCatalyst)
-        if !viewModel.isSearchMode {
-            composerFocused = false
-        }
+        composerFocused = false
         #endif
 
         Task {

ios/Packages/Sybil/Tests/SybilTests/SybilTests.swift

@@ -6,8 +6,20 @@ import Testing
 private struct MockClientCallSnapshot: Sendable {
     var listChats = 0
     var listSearches = 0
+    var createChat = 0
     var getChat = 0
     var getSearch = 0
+    var getActiveRuns = 0
+    var runCompletionStream = 0
+    var attachCompletionStream = 0
+    var attachSearchStream = 0
+}
+
+private struct ChatCreateCallSnapshot: Sendable {
+    var title: String?
+    var provider: Provider?
+    var model: String?
+    var messages: [CompletionRequestMessage]?
 }
 
 private struct UnexpectedClientCall: Error {}
@@ -18,38 +30,68 @@ private actor MockSybilClient: SybilAPIClienting {
     private let chatDetails: [String: ChatDetail]
     private let searchDetails: [String: SearchDetail]
     private let createChatResponse: ChatSummary?
+    private let activeRunsResponse: ActiveRunsResponse
 
     private var snapshot = MockClientCallSnapshot()
+    private var lastCreateChatCall: ChatCreateCallSnapshot?
+    private var lastCompletionStreamBody: CompletionStreamRequest?
+    private var completionStreamEvents: [CompletionStreamEvent]?
+    private var listChatsDelayNanoseconds: UInt64 = 0
+    private var listSearchesDelayNanoseconds: UInt64 = 0
     private var getChatDelayNanoseconds: UInt64 = 0
     private var getSearchDelayNanoseconds: UInt64 = 0
     private var completionStreamNetworkErrorMessage: String?
     private var completionStreamDelayNanoseconds: UInt64 = 0
+    private var completionAttachEvents: [String: [CompletionStreamEvent]] = [:]
+    private var completionAttachDelayNanoseconds: UInt64 = 0
     private var searchStreamNetworkErrorMessage: String?
     private var searchStreamDelayNanoseconds: UInt64 = 0
+    private var searchAttachEvents: [String: [SearchStreamEvent]] = [:]
+    private var searchAttachDelayNanoseconds: UInt64 = 0
 
     init(
         chatsResponse: [ChatSummary] = [],
         searchesResponse: [SearchSummary] = [],
         chatDetails: [String: ChatDetail] = [:],
         searchDetails: [String: SearchDetail] = [:],
-        createChatResponse: ChatSummary? = nil
+        createChatResponse: ChatSummary? = nil,
+        activeRunsResponse: ActiveRunsResponse = ActiveRunsResponse()
     ) {
         self.chatsResponse = chatsResponse
         self.searchesResponse = searchesResponse
         self.chatDetails = chatDetails
         self.searchDetails = searchDetails
         self.createChatResponse = createChatResponse
+        self.activeRunsResponse = activeRunsResponse
     }
 
     func currentSnapshot() -> MockClientCallSnapshot {
         snapshot
     }
 
+    func currentCreateChatCall() -> ChatCreateCallSnapshot? {
+        lastCreateChatCall
+    }
+
+    func currentCompletionStreamBody() -> CompletionStreamRequest? {
+        lastCompletionStreamBody
+    }
+
+    func setCompletionStreamEvents(_ events: [CompletionStreamEvent], delayNanoseconds: UInt64 = 0) {
+        completionStreamEvents = events
+        completionStreamDelayNanoseconds = delayNanoseconds
+    }
+
     func setCompletionStreamNetworkError(_ message: String, delayNanoseconds: UInt64 = 0) {
         completionStreamNetworkErrorMessage = message
         completionStreamDelayNanoseconds = delayNanoseconds
     }
 
+    func setListDelays(chats: UInt64 = 0, searches: UInt64 = 0) {
+        listChatsDelayNanoseconds = chats
+        listSearchesDelayNanoseconds = searches
+    }
+
     func setGetChatDelay(_ delayNanoseconds: UInt64) {
         getChatDelayNanoseconds = delayNanoseconds
     }
@@ -63,16 +105,49 @@ private actor MockSybilClient: SybilAPIClienting {
         searchStreamDelayNanoseconds = delayNanoseconds
     }
 
+    func setCompletionAttachEvents(
+        chatID: String,
+        events: [CompletionStreamEvent],
+        delayNanoseconds: UInt64 = 0
+    ) {
+        completionAttachEvents[chatID] = events
+        completionAttachDelayNanoseconds = delayNanoseconds
+    }
+
+    func setSearchAttachEvents(
+        searchID: String,
+        events: [SearchStreamEvent],
+        delayNanoseconds: UInt64 = 0
+    ) {
+        searchAttachEvents[searchID] = events
+        searchAttachDelayNanoseconds = delayNanoseconds
+    }
+
     func verifySession() async throws -> AuthSession {
         AuthSession(authenticated: true, mode: "open")
     }
 
     func listChats() async throws -> [ChatSummary] {
         snapshot.listChats += 1
+        if listChatsDelayNanoseconds > 0 {
+            try await Task.sleep(nanoseconds: listChatsDelayNanoseconds)
+        }
         return chatsResponse
     }
 
-    func createChat(title: String?) async throws -> ChatSummary {
+    func createChat(
+        title: String?,
+        provider: Provider?,
+        model: String?,
+        messages: [CompletionRequestMessage]?
+    ) async throws -> ChatSummary {
+        snapshot.createChat += 1
+        lastCreateChatCall = ChatCreateCallSnapshot(
+            title: title,
+            provider: provider,
+            model: model,
+            messages: messages
+        )
         if let createChatResponse {
             return createChatResponse
         }
@@ -100,6 +175,9 @@ private actor MockSybilClient: SybilAPIClienting {
 
     func listSearches() async throws -> [SearchSummary] {
         snapshot.listSearches += 1
+        if listSearchesDelayNanoseconds > 0 {
+            try await Task.sleep(nanoseconds: listSearchesDelayNanoseconds)
+        }
         return searchesResponse
     }
 
@@ -130,19 +208,46 @@ private actor MockSybilClient: SybilAPIClienting {
         ModelCatalogResponse(providers: [:])
     }
 
+    func getActiveRuns() async throws -> ActiveRunsResponse {
+        snapshot.getActiveRuns += 1
+        return activeRunsResponse
+    }
+
     func runCompletionStream(
         body: CompletionStreamRequest,
         onEvent: @escaping @Sendable (CompletionStreamEvent) async -> Void
     ) async throws {
+        snapshot.runCompletionStream += 1
+        lastCompletionStreamBody = body
         if completionStreamDelayNanoseconds > 0 {
             try await Task.sleep(nanoseconds: completionStreamDelayNanoseconds)
         }
         if let completionStreamNetworkErrorMessage {
             throw APIError.networkError(message: completionStreamNetworkErrorMessage)
         }
+        if let completionStreamEvents {
+            for event in completionStreamEvents {
+                await onEvent(event)
+            }
+            return
+        }
         throw UnexpectedClientCall()
     }
 
+    func attachCompletionStream(
+        chatID: String,
+        onEvent: @escaping @Sendable (CompletionStreamEvent) async -> Void
+    ) async throws {
+        snapshot.attachCompletionStream += 1
+        let events = completionAttachEvents[chatID] ?? []
+        for event in events {
+            await onEvent(event)
+        }
+        if completionAttachDelayNanoseconds > 0 {
+            try await Task.sleep(nanoseconds: completionAttachDelayNanoseconds)
+        }
+    }
+
     func runSearchStream(
         searchID: String,
         body: SearchRunRequest,
@@ -156,6 +261,20 @@ private actor MockSybilClient: SybilAPIClienting {
         }
         throw UnexpectedClientCall()
     }
+
+    func attachSearchStream(
+        searchID: String,
+        onEvent: @escaping @Sendable (SearchStreamEvent) async -> Void
+    ) async throws {
+        snapshot.attachSearchStream += 1
+        let events = searchAttachEvents[searchID] ?? []
+        for event in events {
+            await onEvent(event)
+        }
+        if searchAttachDelayNanoseconds > 0 {
+            try await Task.sleep(nanoseconds: searchAttachDelayNanoseconds)
+        }
+    }
 }
 
 @MainActor
@@ -277,6 +396,33 @@ private func makeSearchDetail(id: String, date: Date, answer: String) -> SearchD
     #expect(viewModel.selectedItem == .chat("chat-1"))
 }
 
+@MainActor
+@Test func pullToRefreshCompletesWhenRefreshableTaskIsCancelled() async throws {
+    let date = Date(timeIntervalSince1970: 1_700_000_050)
+    let chat = makeChatSummary(id: "chat-cancelled", date: date)
+    let search = makeSearchSummary(id: "search-cancelled", date: date)
+    let client = MockSybilClient(
+        chatsResponse: [chat],
+        searchesResponse: [search]
+    )
+    await client.setListDelays(chats: 50_000_000, searches: 50_000_000)
+    let viewModel = SybilViewModel(settings: testSettings(named: #function)) { _ in client }
+    viewModel.isAuthenticated = true
+    viewModel.isCheckingSession = false
+
+    let refreshTask = Task {
+        await viewModel.refreshSidebarCollectionsFromPullToRefresh()
+    }
+    try await Task.sleep(nanoseconds: 10_000_000)
+    refreshTask.cancel()
+    await refreshTask.value
+
+    #expect(viewModel.errorMessage == nil)
+    #expect(!viewModel.isLoadingCollections)
+    #expect(viewModel.chats.map(\.id) == ["chat-cancelled"])
+    #expect(viewModel.searches.map(\.id) == ["search-cancelled"])
+}
+
 @MainActor
 @Test func foregroundChatRefreshReloadsSelectedTranscript() async throws {
     let date = Date(timeIntervalSince1970: 1_700_000_100)
@@ -409,6 +555,170 @@ private func makeSearchDetail(id: String, date: Date, answer: String) -> SearchD
     await sendTask.value
 }
 
+@MainActor
+@Test func quickQuestionRunsNonPersistentCompletionStream() async throws {
+    let client = MockSybilClient()
+    await client.setCompletionStreamEvents([
+        .delta(CompletionStreamDelta(text: "Reset it from ")),
+        .done(CompletionStreamDone(text: "Reset it from Settings."))
+    ])
+    let viewModel = SybilViewModel(settings: testSettings(named: #function)) { _ in client }
+    viewModel.isAuthenticated = true
+    viewModel.isCheckingSession = false
+    viewModel.quickQuestionPrompt = "How do I reset my password?"
+
+    let task = viewModel.sendQuickQuestion()
+    await task?.value
+
+    let snapshot = await client.currentSnapshot()
+    let body = await client.currentCompletionStreamBody()
+    #expect(snapshot.runCompletionStream == 1)
+    #expect(body?.persist == false)
+    #expect(body?.chatId == nil)
+    #expect(body?.provider == .openai)
+    #expect(body?.messages.first?.role == .user)
+    #expect(body?.messages.first?.content == "How do I reset my password?")
+    #expect(viewModel.quickQuestionAnswerText == "Reset it from Settings.")
+    #expect(!viewModel.isQuickQuestionSending)
+}
+
+@MainActor
+@Test func quickQuestionConvertCreatesSeededChat() async throws {
+    let date = Date(timeIntervalSince1970: 1_700_000_250)
+    let chat = makeChatSummary(id: "quick-chat", date: date)
+    let detail = ChatDetail(
+        id: chat.id,
+        title: chat.title,
+        createdAt: chat.createdAt,
+        updatedAt: chat.updatedAt,
+        initiatedProvider: .openai,
+        initiatedModel: "gpt-4.1-mini",
+        lastUsedProvider: .openai,
+        lastUsedModel: "gpt-4.1-mini",
+        messages: [
+            Message(id: "quick-user", createdAt: date, role: .user, content: "How do I reset my password?", name: nil),
+            Message(id: "quick-assistant", createdAt: date, role: .assistant, content: "Reset it from Settings.", name: nil)
+        ]
+    )
+    let client = MockSybilClient(
+        chatsResponse: [chat],
+        chatDetails: [chat.id: detail],
+        createChatResponse: chat
+    )
+    let viewModel = SybilViewModel(settings: testSettings(named: #function)) { _ in client }
+    viewModel.isAuthenticated = true
+    viewModel.isCheckingSession = false
+    viewModel.quickQuestionSubmittedPrompt = "How do I reset my password?"
+    viewModel.quickQuestionSubmittedProvider = .openai
+    viewModel.quickQuestionSubmittedModel = "gpt-4.1-mini"
+    viewModel.quickQuestionMessages = [
+        Message(
+            id: "temp-assistant-quick",
+            createdAt: date,
+            role: .assistant,
+            content: "Reset it from Settings.",
+            name: nil
+        )
+    ]
+
+    let didConvert = await viewModel.convertQuickQuestionToChat()
+
+    let snapshot = await client.currentSnapshot()
+    let createCall = await client.currentCreateChatCall()
+    #expect(didConvert)
+    #expect(snapshot.createChat == 1)
+    #expect(createCall?.title == "How do I reset my password?")
+    #expect(createCall?.provider == .openai)
+    #expect(createCall?.model == "gpt-4.1-mini")
+    #expect(createCall?.messages?.map(\.role) == [.user, .assistant])
+    #expect(createCall?.messages?.map(\.content) == ["How do I reset my password?", "Reset it from Settings."])
+    #expect(viewModel.selectedItem == .chat("quick-chat"))
+    #expect(viewModel.quickQuestionPrompt.isEmpty)
+}
+
+@MainActor
+@Test func quickQuestionProviderAndModelSelectionPersistSeparately() async throws {
+    let defaults = UserDefaults(suiteName: #function)!
+    defaults.removePersistentDomain(forName: #function)
+    let settings = SybilSettingsStore(defaults: defaults)
+    settings.apiBaseURL = "http://127.0.0.1:8787"
+    let viewModel = SybilViewModel(settings: settings) { _ in MockSybilClient() }
+    viewModel.modelCatalog = [
+        .openai: ProviderModelInfo(models: ["gpt-4.1-mini", "gpt-4o"], loadedAt: nil, error: nil),
+        .anthropic: ProviderModelInfo(models: ["claude-3-5-sonnet-latest", "claude-3-haiku"], loadedAt: nil, error: nil)
+    ]
+
+    viewModel.setQuickQuestionProvider(.anthropic)
+    viewModel.setQuickQuestionModel("claude-3-haiku")
+
+    #expect(viewModel.quickQuestionProvider == .anthropic)
+    #expect(viewModel.quickQuestionModel == "claude-3-haiku")
+    #expect(settings.preferredProvider == .openai)
+
+    let reloadedSettings = SybilSettingsStore(defaults: defaults)
+    #expect(reloadedSettings.quickQuestionPreferredProvider == .anthropic)
+    #expect(reloadedSettings.quickQuestionPreferredModelByProvider[.anthropic] == "claude-3-haiku")
+    #expect(reloadedSettings.preferredProvider == .openai)
+
+    let reloadedViewModel = SybilViewModel(settings: reloadedSettings) { _ in MockSybilClient() }
+    #expect(reloadedViewModel.quickQuestionProvider == .anthropic)
+    #expect(reloadedViewModel.quickQuestionModel == "claude-3-haiku")
+    #expect(reloadedViewModel.provider == .openai)
+}
+
+@MainActor
+@Test func reconnectAttachesSelectedActiveChatStream() async throws {
+    let date = Date(timeIntervalSince1970: 1_700_000_260)
+    let chat = makeChatSummary(id: "chat-active", date: date)
+    let detail = makeChatDetail(id: "chat-active", date: date, body: "existing transcript")
+    let client = MockSybilClient(
+        chatsResponse: [chat],
+        chatDetails: ["chat-active": detail],
+        activeRunsResponse: ActiveRunsResponse(chats: ["chat-active"])
+    )
+    await client.setCompletionAttachEvents(
+        chatID: "chat-active",
+        events: [.delta(CompletionStreamDelta(text: "streaming"))],
+        delayNanoseconds: 100_000_000
+    )
+    let viewModel = SybilViewModel(settings: testSettings(named: #function)) { _ in client }
+
+    await viewModel.reconnect()
+    try await Task.sleep(nanoseconds: 20_000_000)
+
+    let snapshot = await client.currentSnapshot()
+    #expect(snapshot.getActiveRuns >= 1)
+    #expect(snapshot.attachCompletionStream == 1)
+    #expect(viewModel.sidebarItems.first?.isRunning == true)
+    #expect(viewModel.isSendingVisibleChat)
+    #expect(viewModel.displayedMessages.last?.content == "streaming")
+}
+
+@MainActor
+@Test func activeRunOnDifferentChatDoesNotDisableComposer() async throws {
+    let date = Date(timeIntervalSince1970: 1_700_000_270)
+    let activeChat = makeChatSummary(id: "chat-active", date: date)
+    let idleChat = makeChatSummary(id: "chat-idle", date: date.addingTimeInterval(1))
+    let client = MockSybilClient(
+        chatsResponse: [idleChat, activeChat],
+        chatDetails: [
+            "chat-active": makeChatDetail(id: "chat-active", date: date, body: "active transcript"),
+            "chat-idle": makeChatDetail(id: "chat-idle", date: date, body: "idle transcript")
+        ],
+        activeRunsResponse: ActiveRunsResponse(chats: ["chat-active"])
+    )
+    let viewModel = SybilViewModel(settings: testSettings(named: #function)) { _ in client }
+    viewModel.selectedItem = .chat("chat-idle")
+    viewModel.composer = "new message"
+
+    await viewModel.reconnect()
+
+    #expect(viewModel.selectedItem == .chat("chat-idle"))
+    #expect(viewModel.sidebarItems.first(where: { $0.selection == .chat("chat-active") })?.isRunning == true)
+    #expect(!viewModel.isActiveSelectionSending)
+    #expect(viewModel.canSendComposer)
+}
+
 @MainActor
 @Test func backgroundChatStreamInterruptionIsSuppressedUntilForegroundRefresh() async throws {
     let date = Date(timeIntervalSince1970: 1_700_000_300)

server/README.md

@@ -1,7 +1,7 @@
 # Sybil Server
 
 Backend API for:
-- LLM multiplexer (OpenAI Responses / Anthropic / xAI Chat Completions-compatible Grok)
+- LLM multiplexer (OpenAI Responses / Anthropic / xAI Chat Completions-compatible Grok / Hermes Agent)
 - Personal chat database (chats/messages + LLM call log)
 
 ## Stack
@@ -43,6 +43,9 @@ If `ADMIN_TOKEN` is not set, the server runs in open mode (dev).
 - `OPENAI_API_KEY`
 - `ANTHROPIC_API_KEY`
 - `XAI_API_KEY`
+- `HERMES_AGENT_API_BASE_URL` (`http://127.0.0.1:8642/v1` by default; include the `/v1` suffix)
+- `HERMES_AGENT_API_KEY` (enables the Hermes Agent provider; set to Hermes `API_SERVER_KEY`, or any non-empty value if that local server does not require auth)
+- `HERMES_AGENT_MODEL` (optional fallback/override model id; defaults client-side to `hermes-agent`)
 - `EXA_API_KEY`
 - `CHAT_WEB_SEARCH_ENGINE` (`exa` by default, or `searxng` for chat tool calls only)
 - `SEARXNG_BASE_URL` (required when `CHAT_WEB_SEARCH_ENGINE=searxng`; instance must allow `format=json`)

server/prisma/schema.prisma

@@ -13,6 +13,7 @@ enum Provider {
   openai
   anthropic
   xai
+  hermes_agent @map("hermes-agent")
 }
 
 enum MessageRole {

server/src/active-streams.ts

@@ -0,0 +1,59 @@
+export type SseStreamEvent = {
+  event: string;
+  data: unknown;
+};
+
+type SseStreamListener = (event: SseStreamEvent) => void;
+
+export class ActiveSseStream {
+  private readonly events: SseStreamEvent[] = [];
+  private readonly listeners = new Set<SseStreamListener>();
+  private completed = false;
+  private resolveDone!: () => void;
+
+  readonly done: Promise<void>;
+
+  constructor() {
+    this.done = new Promise((resolve) => {
+      this.resolveDone = resolve;
+    });
+  }
+
+  get isCompleted() {
+    return this.completed;
+  }
+
+  emit(event: string, data: unknown) {
+    if (this.completed) return;
+    const entry = { event, data };
+    this.events.push(entry);
+    for (const listener of this.listeners) {
+      listener(entry);
+    }
+  }
+
+  complete(finalEvent?: SseStreamEvent) {
+    if (this.completed) return;
+    if (finalEvent) {
+      this.emit(finalEvent.event, finalEvent.data);
+    }
+    this.completed = true;
+    this.listeners.clear();
+    this.resolveDone();
+  }
+
+  subscribe(listener: SseStreamListener) {
+    for (const event of this.events) {
+      listener(event);
+    }
+
+    if (this.completed) {
+      return () => {};
+    }
+
+    this.listeners.add(listener);
+    return () => {
+      this.listeners.delete(listener);
+    };
+  }
+}

server/src/env.ts

@@ -11,6 +11,13 @@ const OptionalUrlSchema = z.preprocess(
   z.string().trim().url().optional()
 );
 
+const DEFAULT_HERMES_AGENT_API_BASE_URL = "http://127.0.0.1:8642/v1";
+
+const HermesAgentApiBaseUrlSchema = z.preprocess(
+  (value) => (typeof value === "string" && value.trim() === "" ? undefined : value),
+  z.string().trim().url().default(DEFAULT_HERMES_AGENT_API_BASE_URL)
+);
+
 const ChatWebSearchEngineSchema = z.preprocess(
   (value) => {
     if (typeof value !== "string") return value;
@@ -59,6 +66,9 @@ const EnvSchema = z.object({
   OPENAI_API_KEY: z.string().optional(),
   ANTHROPIC_API_KEY: z.string().optional(),
   XAI_API_KEY: z.string().optional(),
+  HERMES_AGENT_API_BASE_URL: HermesAgentApiBaseUrlSchema,
+  HERMES_AGENT_API_KEY: OptionalTrimmedStringSchema,
+  HERMES_AGENT_MODEL: OptionalTrimmedStringSchema,
   EXA_API_KEY: z.string().optional(),
 
   // Chat-mode web_search tool configuration. Search mode remains Exa-only for now.

server/src/llm/chat-tools.ts

@@ -385,6 +385,10 @@ function normalizeIncomingMessages(messages: ChatMessage[]) {
   return [{ role: "system", content: CHAT_TOOL_SYSTEM_PROMPT }, ...normalized];
 }
 
+function normalizePlainIncomingMessages(messages: ChatMessage[]) {
+  return messages.map((message) => buildOpenAIConversationMessage(message));
+}
+
 function normalizeIncomingResponsesInput(messages: ChatMessage[]) {
   const normalized = messages.map((message) => buildOpenAIResponsesInputMessage(message));
 
@@ -853,6 +857,20 @@ function extractResponsesText(response: any, fallback = "") {
   return parts.join("") || fallback;
 }
 
+function extractChatCompletionContent(message: any) {
+  if (typeof message?.content === "string") return message.content;
+  if (!Array.isArray(message?.content)) return "";
+
+  return message.content
+    .map((part: any) => {
+      if (typeof part === "string") return part;
+      if (typeof part?.text === "string") return part.text;
+      if (typeof part?.content === "string") return part.content;
+      return "";
+    })
+    .join("");
+}
+
 function getUnstreamedText(finalText: string, streamedText: string) {
   if (!finalText) return "";
   if (!streamedText) return finalText;
@@ -1093,6 +1111,26 @@ export async function runToolAwareChatCompletions(params: ToolAwareCompletionPar
   };
 }
 
+export async function runPlainChatCompletions(params: ToolAwareCompletionParams): Promise<ToolAwareCompletionResult> {
+  const completion = await params.client.chat.completions.create({
+    model: params.model,
+    messages: normalizePlainIncomingMessages(params.messages),
+    temperature: params.temperature,
+    max_tokens: params.maxTokens,
+  } as any);
+
+  const usageAcc: Required<ToolAwareUsage> = { inputTokens: 0, outputTokens: 0, totalTokens: 0 };
+  const sawUsage = mergeUsage(usageAcc, completion?.usage);
+  const message = completion?.choices?.[0]?.message;
+
+  return {
+    text: extractChatCompletionContent(message),
+    usage: sawUsage ? usageAcc : undefined,
+    raw: { response: completion, api: "chat.completions" },
+    toolEvents: [],
+  };
+}
+
 export async function* runToolAwareOpenAIChatStream(
   params: ToolAwareCompletionParams
 ): AsyncGenerator<ToolAwareStreamingEvent> {
@@ -1354,3 +1392,41 @@ export async function* runToolAwareChatCompletionsStream(
     },
   };
 }
+
+export async function* runPlainChatCompletionsStream(
+  params: ToolAwareCompletionParams
+): AsyncGenerator<ToolAwareStreamingEvent> {
+  const rawResponses: unknown[] = [];
+  const usageAcc: Required<ToolAwareUsage> = { inputTokens: 0, outputTokens: 0, totalTokens: 0 };
+  let sawUsage = false;
+  let text = "";
+
+  const stream = await params.client.chat.completions.create({
+    model: params.model,
+    messages: normalizePlainIncomingMessages(params.messages),
+    temperature: params.temperature,
+    max_tokens: params.maxTokens,
+    stream: true,
+  } as any);
+
+  for await (const chunk of stream as any as AsyncIterable<any>) {
+    rawResponses.push(chunk);
+    sawUsage = mergeUsage(usageAcc, chunk?.usage) || sawUsage;
+
+    const deltaText = chunk?.choices?.[0]?.delta?.content ?? "";
+    if (typeof deltaText === "string" && deltaText.length) {
+      text += deltaText;
+      yield { type: "delta", text: deltaText };
+    }
+  }
+
+  yield {
+    type: "done",
+    result: {
+      text,
+      usage: sawUsage ? usageAcc : undefined,
+      raw: { streamed: true, responses: rawResponses, api: "chat.completions" },
+      toolEvents: [],
+    },
+  };
+}

server/src/llm/model-catalog.ts

@@ -1,5 +1,6 @@
 import type { FastifyBaseLogger } from "fastify";
-import { anthropicClient, openaiClient, xaiClient } from "./providers.js";
+import { env } from "../env.js";
+import { anthropicClient, hermesAgentClient, isHermesAgentConfigured, openaiClient, xaiClient } from "./providers.js";
 import type { Provider } from "./types.js";
 
 export type ProviderModelSnapshot = {
@@ -8,9 +9,9 @@ export type ProviderModelSnapshot = {
   error: string | null;
 };
 
-export type ModelCatalogSnapshot = Record<Provider, ProviderModelSnapshot>;
+export type ModelCatalogSnapshot = Partial<Record<Provider, ProviderModelSnapshot>>;
 
-const providers: Provider[] = ["openai", "anthropic", "xai"];
+const baseProviders: Provider[] = ["openai", "anthropic", "xai"];
 const MODEL_FETCH_TIMEOUT_MS = 15000;
 
 const modelCatalog: ModelCatalogSnapshot = {
@@ -19,6 +20,10 @@ const modelCatalog: ModelCatalogSnapshot = {
   xai: { models: [], loadedAt: null, error: null },
 };
 
+function getCatalogProviders(): Provider[] {
+  return isHermesAgentConfigured() ? [...baseProviders, "hermes-agent"] : baseProviders;
+}
+
 function uniqSorted(models: string[]) {
   return [...new Set(models.map((value) => value.trim()).filter(Boolean))].sort((a, b) => a.localeCompare(b));
 }
@@ -59,8 +64,15 @@ async function fetchProviderModels(provider: Provider) {
     return uniqSorted(page.data.map((model) => model.id));
   }
 
-  const page = await xaiClient().models.list();
-  return uniqSorted(page.data.map((model) => model.id));
+  if (provider === "xai") {
+    const page = await xaiClient().models.list();
+    return uniqSorted(page.data.map((model) => model.id));
+  }
+
+  const page = await hermesAgentClient().models.list();
+  const models = page.data.map((model) => model.id);
+  if (env.HERMES_AGENT_MODEL) models.push(env.HERMES_AGENT_MODEL);
+  return uniqSorted(models);
 }
 
 async function refreshProviderModels(provider: Provider, logger?: FastifyBaseLogger) {
@@ -75,7 +87,7 @@ async function refreshProviderModels(provider: Provider, logger?: FastifyBaseLog
   } catch (err: any) {
     const message = err?.message ?? String(err);
     modelCatalog[provider] = {
-      models: [],
+      models: provider === "hermes-agent" && env.HERMES_AGENT_MODEL ? [env.HERMES_AGENT_MODEL] : [],
       loadedAt: new Date().toISOString(),
       error: message,
     };
@@ -84,25 +96,18 @@ async function refreshProviderModels(provider: Provider, logger?: FastifyBaseLog
 }
 
 export async function warmModelCatalog(logger?: FastifyBaseLogger) {
-  await Promise.all(providers.map((provider) => refreshProviderModels(provider, logger)));
+  await Promise.all(getCatalogProviders().map((provider) => refreshProviderModels(provider, logger)));
 }
 
 export function getModelCatalogSnapshot(): ModelCatalogSnapshot {
-  return {
-    openai: {
-      models: [...modelCatalog.openai.models],
-      loadedAt: modelCatalog.openai.loadedAt,
-      error: modelCatalog.openai.error,
-    },
-    anthropic: {
-      models: [...modelCatalog.anthropic.models],
-      loadedAt: modelCatalog.anthropic.loadedAt,
-      error: modelCatalog.anthropic.error,
-    },
-    xai: {
-      models: [...modelCatalog.xai.models],
-      loadedAt: modelCatalog.xai.loadedAt,
-      error: modelCatalog.xai.error,
-    },
-  };
+  const snapshot: ModelCatalogSnapshot = {};
+  for (const provider of getCatalogProviders()) {
+    const entry = modelCatalog[provider] ?? { models: [], loadedAt: null, error: null };
+    snapshot[provider] = {
+      models: [...entry.models],
+      loadedAt: entry.loadedAt,
+      error: entry.error,
+    };
+  }
+  return snapshot;
 }

server/src/llm/multiplexer.ts

@@ -1,13 +1,13 @@
 import { performance } from "node:perf_hooks";
 import { prisma } from "../db.js";
-import { anthropicClient, openaiClient, xaiClient } from "./providers.js";
-import { buildToolLogMessageData, runToolAwareChatCompletions, runToolAwareOpenAIChat } from "./chat-tools.js";
+import { anthropicClient, hermesAgentClient, openaiClient, xaiClient } from "./providers.js";
+import { buildToolLogMessageData, runPlainChatCompletions, runToolAwareChatCompletions, runToolAwareOpenAIChat } from "./chat-tools.js";
 import { buildAnthropicConversationMessage, getAnthropicSystemPrompt } from "./message-content.js";
+import { toPrismaProvider } from "./provider-ids.js";
 import type { MultiplexRequest, MultiplexResponse, Provider } from "./types.js";
 
 function asProviderEnum(p: Provider) {
-  // Prisma enum values match these strings.
-  return p;
+  return toPrismaProvider(p);
 }
 
 export async function runMultiplex(req: MultiplexRequest): Promise<MultiplexResponse> {
@@ -84,6 +84,23 @@ export async function runMultiplex(req: MultiplexRequest): Promise<MultiplexResp
       outText = r.text;
       usage = r.usage;
       toolMessages = r.toolEvents.map((event) => buildToolLogMessageData(call.chatId, event));
+    } else if (req.provider === "hermes-agent") {
+      const client = hermesAgentClient();
+      const r = await runPlainChatCompletions({
+        client,
+        model: req.model,
+        messages: req.messages,
+        temperature: req.temperature,
+        maxTokens: req.maxTokens,
+        logContext: {
+          provider: req.provider,
+          model: req.model,
+          chatId,
+        },
+      });
+      raw = r.raw;
+      outText = r.text;
+      usage = r.usage;
     } else if (req.provider === "anthropic") {
       const client = anthropicClient();
 

server/src/llm/provider-ids.ts

@@ -0,0 +1,31 @@
+import type { Provider } from "./types.js";
+
+type PrismaProvider = Exclude<Provider, "hermes-agent"> | "hermes_agent";
+
+export function toPrismaProvider(provider: Provider): PrismaProvider {
+  return provider === "hermes-agent" ? "hermes_agent" : provider;
+}
+
+export function fromPrismaProvider(provider: unknown): Provider | null {
+  if (provider === null || provider === undefined) return null;
+  if (provider === "hermes_agent" || provider === "hermes-agent") return "hermes-agent";
+  if (provider === "openai" || provider === "anthropic" || provider === "xai") return provider;
+  return null;
+}
+
+export function serializeProviderFields<T extends Record<string, any>>(value: T): T {
+  const next: Record<string, any> = { ...value };
+  if ("initiatedProvider" in next) {
+    next.initiatedProvider = fromPrismaProvider(next.initiatedProvider);
+  }
+  if ("lastUsedProvider" in next) {
+    next.lastUsedProvider = fromPrismaProvider(next.lastUsedProvider);
+  }
+  if ("provider" in next) {
+    next.provider = fromPrismaProvider(next.provider);
+  }
+  if (Array.isArray(next.calls)) {
+    next.calls = next.calls.map((call: Record<string, any>) => serializeProviderFields(call));
+  }
+  return next as T;
+}

server/src/llm/providers.ts

@@ -13,6 +13,18 @@ export function xaiClient() {
   return new OpenAI({ apiKey: env.XAI_API_KEY, baseURL: "https://api.x.ai/v1" });
 }
 
+export function isHermesAgentConfigured() {
+  return Boolean(env.HERMES_AGENT_API_KEY);
+}
+
+export function hermesAgentClient() {
+  if (!env.HERMES_AGENT_API_KEY) throw new Error("HERMES_AGENT_API_KEY not set");
+  return new OpenAI({
+    apiKey: env.HERMES_AGENT_API_KEY,
+    baseURL: env.HERMES_AGENT_API_BASE_URL,
+  });
+}
+
 export function anthropicClient() {
   if (!env.ANTHROPIC_API_KEY) throw new Error("ANTHROPIC_API_KEY not set");
   return new Anthropic({ apiKey: env.ANTHROPIC_API_KEY });

server/src/llm/streaming.ts

@@ -1,13 +1,15 @@
 import { performance } from "node:perf_hooks";
 import { prisma } from "../db.js";
-import { anthropicClient, openaiClient, xaiClient } from "./providers.js";
+import { anthropicClient, hermesAgentClient, openaiClient, xaiClient } from "./providers.js";
 import {
   buildToolLogMessageData,
+  runPlainChatCompletionsStream,
   runToolAwareChatCompletionsStream,
   runToolAwareOpenAIChatStream,
   type ToolExecutionEvent,
 } from "./chat-tools.js";
 import { buildAnthropicConversationMessage, getAnthropicSystemPrompt } from "./message-content.js";
+import { toPrismaProvider } from "./provider-ids.js";
 import type { MultiplexRequest, Provider } from "./types.js";
 
 type StreamUsage = {
@@ -38,7 +40,7 @@ export async function* runMultiplexStream(req: MultiplexRequest): AsyncGenerator
       ? await prisma.llmCall.create({
           data: {
             chatId,
-            provider: req.provider as any,
+            provider: toPrismaProvider(req.provider) as any,
             model: req.model,
             request: req as any,
           },
@@ -51,14 +53,14 @@ export async function* runMultiplexStream(req: MultiplexRequest): AsyncGenerator
       prisma.chat.update({
         where: { id: chatId },
         data: {
-          lastUsedProvider: req.provider as any,
+          lastUsedProvider: toPrismaProvider(req.provider) as any,
           lastUsedModel: req.model,
         },
       }),
       prisma.chat.updateMany({
         where: { id: chatId, initiatedProvider: null },
         data: {
-          initiatedProvider: req.provider as any,
+          initiatedProvider: toPrismaProvider(req.provider) as any,
           initiatedModel: req.model,
         },
       }),
@@ -72,8 +74,8 @@ export async function* runMultiplexStream(req: MultiplexRequest): AsyncGenerator
   let raw: unknown = { streamed: true };
 
   try {
-    if (req.provider === "openai" || req.provider === "xai") {
-      const client = req.provider === "openai" ? openaiClient() : xaiClient();
+    if (req.provider === "openai" || req.provider === "xai" || req.provider === "hermes-agent") {
+      const client = req.provider === "openai" ? openaiClient() : req.provider === "xai" ? xaiClient() : hermesAgentClient();
       const streamEvents =
         req.provider === "openai"
           ? runToolAwareOpenAIChatStream({
@@ -88,6 +90,19 @@ export async function* runMultiplexStream(req: MultiplexRequest): AsyncGenerator
                 chatId: chatId ?? undefined,
               },
             })
+          : req.provider === "hermes-agent"
+            ? runPlainChatCompletionsStream({
+                client,
+                model: req.model,
+                messages: req.messages,
+                temperature: req.temperature,
+                maxTokens: req.maxTokens,
+                logContext: {
+                  provider: req.provider,
+                  model: req.model,
+                  chatId: chatId ?? undefined,
+                },
+              })
           : runToolAwareChatCompletionsStream({
               client,
               model: req.model,

server/src/llm/types.ts

@@ -1,4 +1,6 @@
-export type Provider = "openai" | "anthropic" | "xai";
+export const PROVIDERS = ["openai", "anthropic", "xai", "hermes-agent"] as const;
+
+export type Provider = (typeof PROVIDERS)[number];
 
 export type ChatImageAttachment = {
   kind: "image";

server/src/routes.ts

@@ -1,17 +1,21 @@
 import { performance } from "node:perf_hooks";
 import { z } from "zod";
-import type { FastifyInstance } from "fastify";
+import type { FastifyInstance, FastifyReply, FastifyRequest } from "fastify";
+import { ActiveSseStream, type SseStreamEvent } from "./active-streams.js";
 import { prisma } from "./db.js";
 import { requireAdmin } from "./auth.js";
 import { env } from "./env.js";
 import { buildComparableAttachments } from "./llm/message-content.js";
 import { runMultiplex } from "./llm/multiplexer.js";
-import { runMultiplexStream } from "./llm/streaming.js";
+import { runMultiplexStream, type StreamEvent } from "./llm/streaming.js";
 import { getModelCatalogSnapshot } from "./llm/model-catalog.js";
 import { openaiClient } from "./llm/providers.js";
+import { serializeProviderFields, toPrismaProvider } from "./llm/provider-ids.js";
 import { exaClient } from "./search/exa.js";
 import type { ChatAttachment } from "./llm/types.js";
 
+const ProviderSchema = z.enum(["openai", "anthropic", "xai", "hermes-agent"]);
+
 type IncomingChatMessage = {
   role: "system" | "user" | "assistant" | "tool";
   content: string;
@@ -120,6 +124,26 @@ const CompletionMessageSchema = z
     }
   });
 
+const CompletionStreamBody = z
+  .object({
+    chatId: z.string().optional(),
+    persist: z.boolean().optional(),
+    provider: ProviderSchema,
+    model: z.string().min(1),
+    messages: z.array(CompletionMessageSchema),
+    temperature: z.number().min(0).max(2).optional(),
+    maxTokens: z.number().int().positive().optional(),
+  })
+  .superRefine((value, ctx) => {
+    if (value.persist === false && value.chatId) {
+      ctx.addIssue({
+        code: z.ZodIssueCode.custom,
+        message: "chatId must be omitted when persist is false",
+        path: ["chatId"],
+      });
+    }
+  });
+
 function mergeAttachmentsIntoMetadata(metadata: unknown, attachments?: ChatAttachment[]) {
   if (!attachments?.length) return metadata as any;
   if (!metadata || typeof metadata !== "object" || Array.isArray(metadata)) {
@@ -293,6 +317,246 @@ function buildSseHeaders(originHeader: string | undefined) {
   return headers;
 }
 
+type SearchRunRequest = z.infer<typeof SearchRunBody>;
+
+const activeChatStreams = new Map<string, ActiveSseStream>();
+const activeSearchStreams = new Map<string, ActiveSseStream>();
+
+function getErrorMessage(err: unknown) {
+  return err instanceof Error ? err.message : String(err);
+}
+
+function writeSseEvent(reply: FastifyReply, event: SseStreamEvent) {
+  if (reply.raw.destroyed || reply.raw.writableEnded) return;
+  reply.raw.write(`event: ${event.event}\n`);
+  reply.raw.write(`data: ${JSON.stringify(event.data)}\n\n`);
+}
+
+async function streamActiveRun(req: FastifyRequest, reply: FastifyReply, stream: ActiveSseStream) {
+  reply.raw.writeHead(200, buildSseHeaders(typeof req.headers.origin === "string" ? req.headers.origin : undefined));
+  reply.raw.flushHeaders?.();
+
+  let unsubscribe = () => {};
+  let closed = false;
+  const closedPromise = new Promise<void>((resolve) => {
+    const onClose = () => {
+      closed = true;
+      unsubscribe();
+      reply.raw.off("close", onClose);
+      resolve();
+    };
+    reply.raw.on("close", onClose);
+    stream.done.finally(() => {
+      reply.raw.off("close", onClose);
+    });
+  });
+
+  unsubscribe = stream.subscribe((event) => writeSseEvent(reply, event));
+  await Promise.race([stream.done, closedPromise]);
+  unsubscribe();
+
+  if (!closed && !reply.raw.destroyed && !reply.raw.writableEnded) {
+    reply.raw.end();
+  }
+
+  return reply;
+}
+
+function mapChatStreamEvent(ev: StreamEvent): SseStreamEvent {
+  if (ev.type === "tool_call") return { event: "tool_call", data: ev.event };
+  return { event: ev.type, data: ev };
+}
+
+function startActiveChatStream(chatId: string, body: z.infer<typeof CompletionStreamBody>) {
+  const stream = new ActiveSseStream();
+  activeChatStreams.set(chatId, stream);
+
+  void (async () => {
+    let sawTerminalEvent = false;
+    try {
+      for await (const ev of runMultiplexStream(body)) {
+        const event = mapChatStreamEvent(ev);
+        if (ev.type === "done" || ev.type === "error") {
+          sawTerminalEvent = true;
+          stream.complete(event);
+          break;
+        }
+        stream.emit(event.event, event.data);
+      }
+
+      if (!sawTerminalEvent) {
+        stream.complete({ event: "error", data: { message: "chat stream ended unexpectedly" } });
+      }
+    } catch (err) {
+      stream.complete({ event: "error", data: { message: getErrorMessage(err) } });
+    } finally {
+      activeChatStreams.delete(chatId);
+    }
+  })();
+
+  return stream;
+}
+
+async function executeSearchRunStream(searchId: string, body: SearchRunRequest, stream: ActiveSseStream) {
+  const startedAt = performance.now();
+  const query = body.query?.trim();
+  if (!query) {
+    stream.complete({ event: "error", data: { message: "query is required" } });
+    return;
+  }
+
+  const normalizedTitle = body.title?.trim() || query.slice(0, 80);
+
+  try {
+    const exa = exaClient();
+    const searchPromise = exa.search(query, {
+      type: body.type ?? "auto",
+      numResults: body.numResults ?? 10,
+      includeDomains: body.includeDomains,
+      excludeDomains: body.excludeDomains,
+      moderation: true,
+      userLocation: "US",
+      contents: false,
+    } as any);
+    const answerPromise = exa.answer(query, {
+      text: true,
+      model: "exa",
+      userLocation: "US",
+    });
+
+    let searchResponse: any | null = null;
+    let answerResponse: any | null = null;
+    let enrichedResults: any[] | null = null;
+    let searchError: string | null = null;
+    let answerError: string | null = null;
+
+    const searchSettled = searchPromise.then(
+      async (value) => {
+        searchResponse = value;
+        const previewResults = (value?.results ?? []).map((result: any, index: number) => mapSearchResultPreview(result, index));
+        stream.emit("search_results", {
+          requestId: value?.requestId ?? null,
+          results: previewResults,
+        });
+
+        const urls = (value?.results ?? []).map((result: any) => result?.url).filter((url: string | undefined) => typeof url === "string");
+        if (!urls.length) return;
+
+        try {
+          const contentsResponse = await exa.getContents(urls, {
+            text: { maxCharacters: 1200 },
+            highlights: {
+              query,
+              maxCharacters: 320,
+              numSentences: 2,
+              highlightsPerUrl: 2,
+            },
+          } as any);
+          const byUrl = new Map<string, any>();
+          for (const contentItem of contentsResponse?.results ?? []) {
+            byUrl.set(normalizeUrlForMatch(contentItem?.url), contentItem);
+          }
+
+          enrichedResults = (value?.results ?? []).map((result: any) => {
+            const contentItem = byUrl.get(normalizeUrlForMatch(result?.url));
+            if (!contentItem) return result;
+            return {
+              ...result,
+              text: contentItem.text ?? result.text ?? null,
+              highlights: Array.isArray(contentItem.highlights) ? contentItem.highlights : result.highlights ?? null,
+              highlightScores: Array.isArray(contentItem.highlightScores) ? contentItem.highlightScores : result.highlightScores ?? null,
+            };
+          });
+
+          stream.emit("search_results", {
+            requestId: value?.requestId ?? null,
+            results: enrichedResults.map((result: any, index: number) => mapSearchResultPreview(result, index)),
+          });
+        } catch {
+          // keep preview results if content enrichment fails
+        }
+      },
+      (reason) => {
+        searchError = reason?.message ?? String(reason);
+        stream.emit("search_error", { error: searchError });
+      }
+    );
+
+    const answerSettled = answerPromise.then(
+      (value) => {
+        answerResponse = value;
+        stream.emit("answer", {
+          answerText: parseAnswerText(value),
+          answerRequestId: value?.requestId ?? null,
+          answerCitations: (value?.citations as any) ?? null,
+        });
+      },
+      (reason) => {
+        answerError = reason?.message ?? String(reason);
+        stream.emit("answer_error", { error: answerError });
+      }
+    );
+
+    await Promise.all([searchSettled, answerSettled]);
+
+    const latencyMs = Math.round(performance.now() - startedAt);
+    const persistedResults = enrichedResults ?? searchResponse?.results ?? [];
+    const rows = persistedResults.map((result: any, index: number) => mapSearchResultRow(searchId, result, index));
+    const answerText = parseAnswerText(answerResponse);
+
+    await prisma.$transaction(async (tx) => {
+      await tx.search.update({
+        where: { id: searchId },
+        data: {
+          query,
+          title: normalizedTitle,
+          requestId: searchResponse?.requestId ?? null,
+          rawResponse: searchResponse as any,
+          latencyMs,
+          error: searchError,
+          answerText,
+          answerRequestId: answerResponse?.requestId ?? null,
+          answerCitations: (answerResponse?.citations as any) ?? null,
+          answerRawResponse: answerResponse as any,
+          answerError,
+        },
+      });
+      await tx.searchResult.deleteMany({ where: { searchId } });
+      if (rows.length) {
+        await tx.searchResult.createMany({ data: rows as any });
+      }
+    });
+
+    const search = await prisma.search.findUnique({
+      where: { id: searchId },
+      include: { results: { orderBy: { rank: "asc" } } },
+    });
+    if (!search) {
+      stream.complete({ event: "error", data: { message: "search not found" } });
+    } else {
+      stream.complete({ event: "done", data: { search } });
+    }
+  } catch (err) {
+    const message = getErrorMessage(err);
+    try {
+      await prisma.search.update({
+        where: { id: searchId },
+        data: {
+          query,
+          title: normalizedTitle,
+          latencyMs: Math.round(performance.now() - startedAt),
+          error: message,
+        },
+      });
+    } catch {
+      // keep the stream terminal event even if the backing search row disappeared
+    }
+    stream.complete({ event: "error", data: { message } });
+  } finally {
+    activeSearchStreams.delete(searchId);
+  }
+}
+
 export async function registerRoutes(app: FastifyInstance) {
   app.get("/health", { logLevel: "silent" }, async () => ({ ok: true }));
 
@@ -306,6 +570,14 @@ export async function registerRoutes(app: FastifyInstance) {
     return { providers: getModelCatalogSnapshot() };
   });
 
+  app.get("/v1/active-runs", async (req) => {
+    requireAdmin(req);
+    return {
+      chats: Array.from(activeChatStreams.keys()),
+      searches: Array.from(activeSearchStreams.keys()),
+    };
+  });
+
   app.get("/v1/chats", async (req) => {
     requireAdmin(req);
     const chats = await prisma.chat.findMany({
@@ -322,7 +594,7 @@ export async function registerRoutes(app: FastifyInstance) {
         lastUsedModel: true,
       },
     });
-    return { chats };
+    return { chats: chats.map((chat) => serializeProviderFields(chat)) };
   });
 
   app.post("/v1/chats", async (req) => {
@@ -330,7 +602,7 @@ export async function registerRoutes(app: FastifyInstance) {
     const Body = z
       .object({
         title: z.string().optional(),
-        provider: z.enum(["openai", "anthropic", "xai"]).optional(),
+        provider: ProviderSchema.optional(),
         model: z.string().trim().min(1).optional(),
         messages: z.array(CompletionMessageSchema).optional(),
       })
@@ -356,9 +628,9 @@ export async function registerRoutes(app: FastifyInstance) {
     const chat = await prisma.chat.create({
       data: {
         title: body.title,
-        initiatedProvider: body.provider as any,
+        initiatedProvider: body.provider ? (toPrismaProvider(body.provider) as any) : undefined,
         initiatedModel: body.model,
-        lastUsedProvider: body.provider as any,
+        lastUsedProvider: body.provider ? (toPrismaProvider(body.provider) as any) : undefined,
         lastUsedModel: body.model,
         messages: body.messages?.length
           ? {
@@ -382,7 +654,7 @@ export async function registerRoutes(app: FastifyInstance) {
         lastUsedModel: true,
       },
     });
-    return { chat };
+    return { chat: serializeProviderFields(chat) };
   });
 
   app.patch("/v1/chats/:chatId", async (req) => {
@@ -413,7 +685,7 @@ export async function registerRoutes(app: FastifyInstance) {
       },
     });
     if (!chat) return app.httpErrors.notFound("chat not found");
-    return { chat };
+    return { chat: serializeProviderFields(chat) };
   });
 
   app.post("/v1/chats/title/suggest", async (req) => {
@@ -438,7 +710,7 @@ export async function registerRoutes(app: FastifyInstance) {
       },
     });
     if (!existing) return app.httpErrors.notFound("chat not found");
-    if (existing.title?.trim()) return { chat: existing };
+    if (existing.title?.trim()) return { chat: serializeProviderFields(existing) };
 
     const fallback = body.content.split(/\r?\n/)[0]?.trim().slice(0, 48) || "New chat";
     const suggestedRaw = await generateChatTitle(body.content);
@@ -459,7 +731,7 @@ export async function registerRoutes(app: FastifyInstance) {
       },
     });
 
-    return { chat };
+    return { chat: serializeProviderFields(chat) };
   });
 
   app.delete("/v1/chats/:chatId", async (req) => {
@@ -579,7 +851,7 @@ export async function registerRoutes(app: FastifyInstance) {
       },
     });
 
-    return { chat };
+    return { chat: serializeProviderFields(chat) };
   });
 
   app.post("/v1/searches/:searchId/run", async (req) => {
@@ -695,162 +967,24 @@ export async function registerRoutes(app: FastifyInstance) {
     const query = body.query?.trim() || existing.query?.trim();
     if (!query) return app.httpErrors.badRequest("query is required");
 
-    const startedAt = performance.now();
-    const normalizedTitle = body.title?.trim() || query.slice(0, 80);
-
-    reply.raw.writeHead(200, buildSseHeaders(typeof req.headers.origin === "string" ? req.headers.origin : undefined));
-
-    const send = (event: string, data: any) => {
-      if (reply.raw.writableEnded) return;
-      reply.raw.write(`event: ${event}\n`);
-      reply.raw.write(`data: ${JSON.stringify(data)}\n\n`);
-    };
-
-    try {
-      const exa = exaClient();
-      const searchPromise = exa.search(query, {
-        type: body.type ?? "auto",
-        numResults: body.numResults ?? 10,
-        includeDomains: body.includeDomains,
-        excludeDomains: body.excludeDomains,
-        moderation: true,
-        userLocation: "US",
-        contents: false,
-      } as any);
-      const answerPromise = exa.answer(query, {
-        text: true,
-        model: "exa",
-        userLocation: "US",
-      });
-
-      let searchResponse: any | null = null;
-      let answerResponse: any | null = null;
-      let enrichedResults: any[] | null = null;
-      let searchError: string | null = null;
-      let answerError: string | null = null;
-
-      const searchSettled = searchPromise.then(
-        async (value) => {
-          searchResponse = value;
-          const previewResults = (value?.results ?? []).map((result: any, index: number) => mapSearchResultPreview(result, index));
-          send("search_results", {
-            requestId: value?.requestId ?? null,
-            results: previewResults,
-          });
-
-          const urls = (value?.results ?? []).map((result: any) => result?.url).filter((url: string | undefined) => typeof url === "string");
-          if (!urls.length) return;
-
-          try {
-            const contentsResponse = await exa.getContents(urls, {
-              text: { maxCharacters: 1200 },
-              highlights: {
-                query,
-                maxCharacters: 320,
-                numSentences: 2,
-                highlightsPerUrl: 2,
-              },
-            } as any);
-            const byUrl = new Map<string, any>();
-            for (const contentItem of contentsResponse?.results ?? []) {
-              byUrl.set(normalizeUrlForMatch(contentItem?.url), contentItem);
-            }
-
-            enrichedResults = (value?.results ?? []).map((result: any) => {
-              const contentItem = byUrl.get(normalizeUrlForMatch(result?.url));
-              if (!contentItem) return result;
-              return {
-                ...result,
-                text: contentItem.text ?? result.text ?? null,
-                highlights: Array.isArray(contentItem.highlights) ? contentItem.highlights : result.highlights ?? null,
-                highlightScores: Array.isArray(contentItem.highlightScores) ? contentItem.highlightScores : result.highlightScores ?? null,
-              };
-            });
-
-            send("search_results", {
-              requestId: value?.requestId ?? null,
-              results: enrichedResults.map((result: any, index: number) => mapSearchResultPreview(result, index)),
-            });
-          } catch {
-            // keep preview results if content enrichment fails
-          }
-        },
-        (reason) => {
-          searchError = reason?.message ?? String(reason);
-          send("search_error", { error: searchError });
-        }
-      );
-
-      const answerSettled = answerPromise.then(
-        (value) => {
-          answerResponse = value;
-          send("answer", {
-            answerText: parseAnswerText(value),
-            answerRequestId: value?.requestId ?? null,
-            answerCitations: (value?.citations as any) ?? null,
-          });
-        },
-        (reason) => {
-          answerError = reason?.message ?? String(reason);
-          send("answer_error", { error: answerError });
-        }
-      );
-
-      await Promise.all([searchSettled, answerSettled]);
-
-      const latencyMs = Math.round(performance.now() - startedAt);
-      const persistedResults = enrichedResults ?? searchResponse?.results ?? [];
-      const rows = persistedResults.map((result: any, index: number) => mapSearchResultRow(searchId, result, index));
-      const answerText = parseAnswerText(answerResponse);
-
-      await prisma.$transaction(async (tx) => {
-        await tx.search.update({
-          where: { id: searchId },
-          data: {
-            query,
-            title: normalizedTitle,
-            requestId: searchResponse?.requestId ?? null,
-            rawResponse: searchResponse as any,
-            latencyMs,
-            error: searchError,
-            answerText,
-            answerRequestId: answerResponse?.requestId ?? null,
-            answerCitations: (answerResponse?.citations as any) ?? null,
-            answerRawResponse: answerResponse as any,
-            answerError,
-          },
-        });
-        await tx.searchResult.deleteMany({ where: { searchId } });
-        if (rows.length) {
-          await tx.searchResult.createMany({ data: rows as any });
-        }
-      });
-
-      const search = await prisma.search.findUnique({
-        where: { id: searchId },
-        include: { results: { orderBy: { rank: "asc" } } },
-      });
-      if (!search) {
-        send("error", { message: "search not found" });
-      } else {
-        send("done", { search });
-      }
-    } catch (err: any) {
-      await prisma.search.update({
-        where: { id: searchId },
-        data: {
-          query,
-          title: normalizedTitle,
-          latencyMs: Math.round(performance.now() - startedAt),
-          error: err?.message ?? String(err),
-        },
-      });
-      send("error", { message: err?.message ?? String(err) });
-    } finally {
-      reply.raw.end();
+    const existingStream = activeSearchStreams.get(searchId);
+    if (existingStream) {
+      return streamActiveRun(req, reply, existingStream);
     }
 
-    return reply;
+    const stream = new ActiveSseStream();
+    activeSearchStreams.set(searchId, stream);
+    void executeSearchRunStream(searchId, { ...body, query }, stream);
+    return streamActiveRun(req, reply, stream);
+  });
+
+  app.post("/v1/searches/:searchId/run/stream/attach", async (req, reply) => {
+    requireAdmin(req);
+    const Params = z.object({ searchId: z.string() });
+    const { searchId } = Params.parse(req.params);
+    const stream = activeSearchStreams.get(searchId);
+    if (!stream) return app.httpErrors.notFound("active search stream not found");
+    return streamActiveRun(req, reply, stream);
   });
 
   app.get("/v1/chats/:chatId", async (req) => {
@@ -863,7 +997,7 @@ export async function registerRoutes(app: FastifyInstance) {
       include: { messages: { orderBy: { createdAt: "asc" } }, calls: { orderBy: { createdAt: "desc" } } },
     });
     if (!chat) return app.httpErrors.notFound("chat not found");
-    return { chat };
+    return { chat: serializeProviderFields(chat) };
   });
 
   app.post("/v1/chats/:chatId/messages", async (req) => {
@@ -895,13 +1029,22 @@ export async function registerRoutes(app: FastifyInstance) {
     return { message: msg };
   });
 
+  app.post("/v1/chats/:chatId/stream/attach", async (req, reply) => {
+    requireAdmin(req);
+    const Params = z.object({ chatId: z.string() });
+    const { chatId } = Params.parse(req.params);
+    const stream = activeChatStreams.get(chatId);
+    if (!stream) return app.httpErrors.notFound("active chat stream not found");
+    return streamActiveRun(req, reply, stream);
+  });
+
   // Main: create a completion via provider+model and store everything.
   app.post("/v1/chat-completions", async (req) => {
     requireAdmin(req);
 
     const Body = z.object({
       chatId: z.string().optional(),
-      provider: z.enum(["openai", "anthropic", "xai"]),
+      provider: ProviderSchema,
       model: z.string().min(1),
       messages: z.array(CompletionMessageSchema),
       temperature: z.number().min(0).max(2).optional(),
@@ -935,27 +1078,7 @@ export async function registerRoutes(app: FastifyInstance) {
   app.post("/v1/chat-completions/stream", async (req, reply) => {
     requireAdmin(req);
 
-    const Body = z
-      .object({
-        chatId: z.string().optional(),
-        persist: z.boolean().optional(),
-        provider: z.enum(["openai", "anthropic", "xai"]),
-        model: z.string().min(1),
-        messages: z.array(CompletionMessageSchema),
-        temperature: z.number().min(0).max(2).optional(),
-        maxTokens: z.number().int().positive().optional(),
-      })
-      .superRefine((value, ctx) => {
-        if (value.persist === false && value.chatId) {
-          ctx.addIssue({
-            code: z.ZodIssueCode.custom,
-            message: "chatId must be omitted when persist is false",
-            path: ["chatId"],
-          });
-        }
-      });
-
-    const parsed = Body.safeParse(req.body);
+    const parsed = CompletionStreamBody.safeParse(req.body);
     if (!parsed.success) return app.httpErrors.badRequest(parsed.error.message);
     const body = parsed.data;
 
@@ -970,23 +1093,24 @@ export async function registerRoutes(app: FastifyInstance) {
       await storeNonAssistantMessages(body.chatId, body.messages);
     }
 
+    if (body.persist !== false && body.chatId) {
+      if (activeChatStreams.has(body.chatId)) {
+        return app.httpErrors.conflict("chat completion already running");
+      }
+      const stream = startActiveChatStream(body.chatId, body);
+      return streamActiveRun(req, reply, stream);
+    }
+
     reply.raw.writeHead(200, buildSseHeaders(typeof req.headers.origin === "string" ? req.headers.origin : undefined));
     reply.raw.flushHeaders();
 
-    const send = (event: string, data: any) => {
-      reply.raw.write(`event: ${event}\n`);
-      reply.raw.write(`data: ${JSON.stringify(data)}\n\n`);
-    };
-
     for await (const ev of runMultiplexStream(body)) {
-      if (ev.type === "meta") send("meta", ev);
-      else if (ev.type === "tool_call") send("tool_call", ev.event);
-      else if (ev.type === "delta") send("delta", ev);
-      else if (ev.type === "done") send("done", ev);
-      else if (ev.type === "error") send("error", ev);
+      writeSseEvent(reply, mapChatStreamEvent(ev));
     }
 
-    reply.raw.end();
+    if (!reply.raw.destroyed && !reply.raw.writableEnded) {
+      reply.raw.end();
+    }
     return reply;
   });
 }

server/tests/active-streams.test.ts

@@ -0,0 +1,34 @@
+import assert from "node:assert/strict";
+import test from "node:test";
+import { ActiveSseStream, type SseStreamEvent } from "../src/active-streams.js";
+
+test("ActiveSseStream replays buffered events to late subscribers", () => {
+  const stream = new ActiveSseStream();
+  stream.emit("delta", { text: "hel" });
+  stream.emit("delta", { text: "lo" });
+
+  const events: SseStreamEvent[] = [];
+  const unsubscribe = stream.subscribe((event) => events.push(event));
+  unsubscribe();
+
+  assert.deepEqual(events, [
+    { event: "delta", data: { text: "hel" } },
+    { event: "delta", data: { text: "lo" } },
+  ]);
+});
+
+test("ActiveSseStream replays terminal events after completion", async () => {
+  const stream = new ActiveSseStream();
+  stream.emit("delta", { text: "done" });
+  stream.complete({ event: "done", data: { text: "done" } });
+  await stream.done;
+
+  const events: SseStreamEvent[] = [];
+  stream.subscribe((event) => events.push(event));
+
+  assert.equal(stream.isCompleted, true);
+  assert.deepEqual(events, [
+    { event: "delta", data: { text: "done" } },
+    { event: "done", data: { text: "done" } },
+  ]);
+});

server/tests/chat-tools-streaming.test.ts

@@ -1,6 +1,7 @@
 import assert from "node:assert/strict";
 import test from "node:test";
 import {
+  runPlainChatCompletionsStream,
   runToolAwareChatCompletionsStream,
   runToolAwareOpenAIChatStream,
   type ToolAwareStreamingEvent,
@@ -105,3 +106,37 @@ test("OpenAI-compatible Chat Completions stream emits text deltas as they arrive
   );
   assert.equal(events.at(-1)?.type === "done" ? events.at(-1)?.result.text : null, "Hello");
 });
+
+test("plain Chat Completions stream does not send Sybil-managed tools", async () => {
+  let requestBody: any = null;
+  const client = {
+    chat: {
+      completions: {
+        create: async (body: any) => {
+          requestBody = body;
+          return streamFrom([
+            { choices: [{ delta: { content: "Hi" } }] },
+            { choices: [{ delta: {}, finish_reason: "stop" }] },
+          ]);
+        },
+      },
+    },
+  };
+
+  const events = await collectEvents(
+    runPlainChatCompletionsStream({
+      client: client as any,
+      model: "hermes-agent",
+      messages: [{ role: "user", content: "Say hi" }],
+    })
+  );
+
+  assert.equal(requestBody.model, "hermes-agent");
+  assert.equal(requestBody.stream, true);
+  assert.equal("tools" in requestBody, false);
+  assert.deepEqual(
+    events.map((event) => event.type),
+    ["delta", "done"]
+  );
+  assert.equal(events.at(-1)?.type === "done" ? events.at(-1)?.result.text : null, "Hi");
+});

server/tests/provider-ids.test.ts

@@ -0,0 +1,12 @@
+import assert from "node:assert/strict";
+import test from "node:test";
+import { fromPrismaProvider, serializeProviderFields, toPrismaProvider } from "../src/llm/provider-ids.js";
+
+test("Hermes Agent provider id maps between API and Prisma enum forms", () => {
+  assert.equal(toPrismaProvider("hermes-agent"), "hermes_agent");
+  assert.equal(fromPrismaProvider("hermes_agent"), "hermes-agent");
+  assert.deepEqual(serializeProviderFields({ initiatedProvider: "hermes_agent", lastUsedProvider: "xai" }), {
+    initiatedProvider: "hermes-agent",
+    lastUsedProvider: "xai",
+  });
+});

tui/README.md

@@ -23,7 +23,7 @@ Configuration is environment-only (no in-app settings).
 
 - `SYBIL_TUI_API_BASE_URL`: API base URL. Default: `http://127.0.0.1:8787`
 - `SYBIL_TUI_ADMIN_TOKEN`: optional bearer token for token-mode servers
-- `SYBIL_TUI_DEFAULT_PROVIDER`: `openai` | `anthropic` | `xai` (default: `openai`)
+- `SYBIL_TUI_DEFAULT_PROVIDER`: `openai` | `anthropic` | `xai` | `hermes-agent` (default: `openai`)
 - `SYBIL_TUI_DEFAULT_MODEL`: optional default model name
 - `SYBIL_TUI_SEARCH_NUM_RESULTS`: results per search run (default: `10`)
 

tui/src/config.ts

@@ -1,6 +1,6 @@
 import type { Provider } from "./types.js";
 
-const PROVIDERS: Provider[] = ["openai", "anthropic", "xai"];
+const PROVIDERS: Provider[] = ["openai", "anthropic", "xai", "hermes-agent"];
 
 function normalizeBaseUrl(value: string) {
   const trimmed = value.trim();

tui/src/index.ts

@@ -39,11 +39,13 @@ type ToolLogMetadata = {
   resultPreview?: string | null;
 };
 
-const PROVIDERS: Provider[] = ["openai", "anthropic", "xai"];
+const BASE_PROVIDERS: Provider[] = ["openai", "anthropic", "xai"];
+const PROVIDERS: Provider[] = [...BASE_PROVIDERS, "hermes-agent"];
 const PROVIDER_FALLBACK_MODELS: Record<Provider, string[]> = {
   openai: ["gpt-4.1-mini"],
   anthropic: ["claude-3-5-sonnet-latest"],
   xai: ["grok-3-mini"],
+  "hermes-agent": ["hermes-agent"],
 };
 
 const EMPTY_MODEL_CATALOG: ModelCatalogResponse["providers"] = {
@@ -74,6 +76,7 @@ function getProviderLabel(provider: Provider | null | undefined) {
   if (provider === "openai") return "OpenAI";
   if (provider === "anthropic") return "Anthropic";
   if (provider === "xai") return "xAI";
+  if (provider === "hermes-agent") return "Hermes Agent";
   return "";
 }
 
@@ -159,6 +162,10 @@ function getModelOptions(catalog: ModelCatalogResponse["providers"], provider: P
   return PROVIDER_FALLBACK_MODELS[provider];
 }
 
+function getVisibleProviders(catalog: ModelCatalogResponse["providers"]) {
+  return PROVIDERS.filter((provider) => provider !== "hermes-agent" || catalog[provider] !== undefined);
+}
+
 function pickProviderModel(options: string[], preferred: string | null, fallback: string | null = null) {
   if (fallback && options.includes(fallback)) return fallback;
   if (preferred && options.includes(preferred)) return preferred;
@@ -202,6 +209,7 @@ async function main() {
     openai: null,
     anthropic: null,
     xai: null,
+    "hermes-agent": null,
   };
   let model: string = config.defaultModel ?? pickProviderModel(getModelOptions(modelCatalog, provider), null);
   let errorMessage: string | null = null;
@@ -1257,8 +1265,10 @@ async function main() {
   }
 
   function cycleProvider() {
-    const currentIndex = PROVIDERS.indexOf(provider);
-    const nextProvider: Provider = PROVIDERS[(currentIndex + 1) % PROVIDERS.length] ?? "openai";
+    const visibleProviders = getVisibleProviders(modelCatalog);
+    const cycleProviders = visibleProviders.length ? visibleProviders : BASE_PROVIDERS;
+    const currentIndex = Math.max(0, cycleProviders.indexOf(provider));
+    const nextProvider: Provider = cycleProviders[(currentIndex + 1) % cycleProviders.length] ?? "openai";
     provider = nextProvider;
     syncModelForProvider();
     updateUI();

tui/src/types.ts

@@ -1,4 +1,4 @@
-export type Provider = "openai" | "anthropic" | "xai";
+export type Provider = "openai" | "anthropic" | "xai" | "hermes-agent";
 
 export type ProviderModelInfo = {
   models: string[];
@@ -7,7 +7,7 @@ export type ProviderModelInfo = {
 };
 
 export type ModelCatalogResponse = {
-  providers: Record<Provider, ProviderModelInfo>;
+  providers: Partial<Record<Provider, ProviderModelInfo>>;
 };
 
 export type ChatSummary = {

web/src/index.css

@@ -206,17 +206,31 @@ textarea {
 }
 
 .md-content code {
-  background: hsl(288 22% 23%);
-  border-radius: 0.25rem;
+  background: hsl(249 40% 10% / 0.78);
+  border-radius: 0.3rem;
   padding: 0.05rem 0.3rem;
   font-size: 0.86em;
+  box-decoration-break: clone;
+  -webkit-box-decoration-break: clone;
 }
 
 .md-content pre {
   overflow-x: auto;
-  border-radius: 0.5rem;
-  background: hsl(287 28% 13%);
-  padding: 0.6rem 0.75rem;
+  border: 1px solid hsl(253 31% 29% / 0.72);
+  border-radius: 0.625rem;
+  background: hsl(249 40% 10% / 0.82);
+  padding: 0.75rem;
+  box-shadow: inset 0 1px 0 hsl(258 80% 88% / 0.05);
+}
+
+.md-content pre code {
+  display: block;
+  background: transparent;
+  border-radius: 0;
+  padding: 0;
+  font-size: 0.88em;
+  line-height: 1.55;
+  white-space: pre;
 }
 
 .md-content a {

web/src/lib/api.ts

@@ -127,7 +127,7 @@ export type CompletionRequestMessage = {
   attachments?: ChatAttachment[];
 };
 
-export type Provider = "openai" | "anthropic" | "xai";
+export type Provider = "openai" | "anthropic" | "xai" | "hermes-agent";
 
 export type ProviderModelInfo = {
   models: string[];
@@ -136,7 +136,12 @@ export type ProviderModelInfo = {
 };
 
 export type ModelCatalogResponse = {
-  providers: Record<Provider, ProviderModelInfo>;
+  providers: Partial<Record<Provider, ProviderModelInfo>>;
+};
+
+export type ActiveRunsResponse = {
+  chats: string[];
+  searches: string[];
 };
 
 type CompletionResponse = {
@@ -217,6 +222,10 @@ export async function listModels() {
   return api<ModelCatalogResponse>("/v1/models");
 }
 
+export async function getActiveRuns() {
+  return api<ActiveRunsResponse>("/v1/active-runs");
+}
+
 export async function createChat(input?: string | CreateChatRequest) {
   const body = typeof input === "string" ? { title: input } : input ?? {};
   const data = await api<{ chat: ChatSummary }>("/v1/chats", {
@@ -333,6 +342,85 @@ type RunSearchStreamHandlers = {
   onError?: (payload: { message: string }) => void;
 };
 
+async function readSseStream(response: Response, dispatch: (eventName: string, payload: any) => void) {
+  if (!response.ok) {
+    const fallback = `${response.status} ${response.statusText}`;
+    let message = fallback;
+    try {
+      const body = (await response.json()) as { message?: string };
+      if (body.message) message = body.message;
+    } catch {
+      // keep fallback message
+    }
+    throw new Error(message);
+  }
+
+  if (!response.body) {
+    throw new Error("No response stream");
+  }
+
+  const reader = response.body.getReader();
+  const decoder = new TextDecoder();
+  let buffer = "";
+  let eventName = "message";
+  let dataLines: string[] = [];
+
+  const flushEvent = () => {
+    if (!dataLines.length) {
+      eventName = "message";
+      return;
+    }
+
+    const dataText = dataLines.join("\n");
+    let payload: any = null;
+    try {
+      payload = JSON.parse(dataText);
+    } catch {
+      payload = { message: dataText };
+    }
+
+    dispatch(eventName, payload);
+
+    dataLines = [];
+    eventName = "message";
+  };
+
+  while (true) {
+    const { value, done } = await reader.read();
+    if (done) break;
+
+    buffer += decoder.decode(value, { stream: true });
+    let newlineIndex = buffer.indexOf("\n");
+
+    while (newlineIndex >= 0) {
+      const rawLine = buffer.slice(0, newlineIndex);
+      buffer = buffer.slice(newlineIndex + 1);
+      const line = rawLine.endsWith("\r") ? rawLine.slice(0, -1) : rawLine;
+
+      if (!line) {
+        flushEvent();
+      } else if (line.startsWith("event:")) {
+        eventName = line.slice("event:".length).trim();
+      } else if (line.startsWith("data:")) {
+        dataLines.push(line.slice("data:".length).trimStart());
+      }
+
+      newlineIndex = buffer.indexOf("\n");
+    }
+  }
+
+  buffer += decoder.decode();
+  if (buffer.length) {
+    const line = buffer.endsWith("\r") ? buffer.slice(0, -1) : buffer;
+    if (line.startsWith("event:")) {
+      eventName = line.slice("event:".length).trim();
+    } else if (line.startsWith("data:")) {
+      dataLines.push(line.slice("data:".length).trimStart());
+    }
+  }
+  flushEvent();
+}
+
 export async function runSearchStream(
   searchId: string,
   body: SearchRunRequest,
@@ -437,6 +525,30 @@ export async function runSearchStream(
   flushEvent();
 }
 
+export async function attachSearchStream(searchId: string, handlers: RunSearchStreamHandlers, options?: { signal?: AbortSignal }) {
+  const headers = new Headers({
+    Accept: "text/event-stream",
+  });
+  if (authToken) {
+    headers.set("Authorization", `Bearer ${authToken}`);
+  }
+
+  const response = await fetch(`${API_BASE_URL}/v1/searches/${searchId}/run/stream/attach`, {
+    method: "POST",
+    headers,
+    signal: options?.signal,
+  });
+
+  await readSseStream(response, (eventName, payload) => {
+    if (eventName === "search_results") handlers.onSearchResults?.(payload);
+    else if (eventName === "search_error") handlers.onSearchError?.(payload);
+    else if (eventName === "answer") handlers.onAnswer?.(payload);
+    else if (eventName === "answer_error") handlers.onAnswerError?.(payload);
+    else if (eventName === "done") handlers.onDone?.(payload);
+    else if (eventName === "error") handlers.onError?.(payload);
+  });
+}
+
 export async function runCompletion(body: {
   chatId: string;
   provider: Provider;
@@ -556,3 +668,26 @@ export async function runCompletionStream(
   }
   flushEvent();
 }
+
+export async function attachCompletionStream(chatId: string, handlers: CompletionStreamHandlers, options?: { signal?: AbortSignal }) {
+  const headers = new Headers({
+    Accept: "text/event-stream",
+  });
+  if (authToken) {
+    headers.set("Authorization", `Bearer ${authToken}`);
+  }
+
+  const response = await fetch(`${API_BASE_URL}/v1/chats/${chatId}/stream/attach`, {
+    method: "POST",
+    headers,
+    signal: options?.signal,
+  });
+
+  await readSseStream(response, (eventName, payload) => {
+    if (eventName === "meta") handlers.onMeta?.(payload);
+    else if (eventName === "tool_call") handlers.onToolCall?.(payload);
+    else if (eventName === "delta") handlers.onDelta?.(payload);
+    else if (eventName === "done") handlers.onDone?.(payload);
+    else if (eventName === "error") handlers.onError?.(payload);
+  });
+}