Add backend unit test framework and modularize settings UI

Introduced a Vitest-based backend unit testing framework with supporting scripts, helpers, and GitHub Actions integration. Refactored the admin settings page to a modular architecture, splitting monolithic logic into feature-specific tabs and hooks for improved maintainability and testability. Updated documentation to reflect the new testing setup and settings architecture, and added new dependencies for testing utilities.

documentation/README.md

@@ -66,6 +66,9 @@ External integrations: Plex (auth + library), Prowlarr/Jackett (indexers), qBitt
 **Deployment:**
 - [deployment/docker.md](deployment/docker.md) - Docker Compose, volumes, env vars
 
+**Testing:**
+- [testing.md](testing.md) - Backend unit tests, scripts
+
 ## Development Phases
 ✅ Phase 1: Foundation (auth, database, setup wizard)
 ✅ Phase 2: User features (discovery, requests, dashboard)

documentation/TABLEOFCONTENTS.md

@@ -11,7 +11,8 @@
 ## Configuration & Setup
 - **First-time setup wizard** → [setup-wizard.md](setup-wizard.md)
 - **Settings management, encryption** → [backend/services/config.md](backend/services/config.md)
-- **Settings UI (Plex, Prowlarr, paths)** → [settings-pages.md](settings-pages.md)
+- **Settings UI (modular architecture, all tabs)** → [settings-pages.md](settings-pages.md)
+- **Settings architecture refactoring (Jan 2026)** → [settings-pages.md](settings-pages.md#architecture-refactored-jan-2026)
 - **Setup middleware & status check** → [backend/middleware.md](backend/middleware.md)
 - **Environment variables, PUBLIC_URL, OAuth configuration** → [backend/services/environment.md](backend/services/environment.md)
 
@@ -81,6 +82,9 @@
 - **Environment variables, volumes** → [deployment/docker.md](deployment/docker.md)
 - **Database setup (Prisma), migrations** → [deployment/docker.md](deployment/docker.md)
 
+## Testing
+- **Backend unit test framework, scripts** [testing.md](testing.md)
+
 ## Feature-Specific Lookups
 **"How do I add a new audiobook?"** → [integrations/audible.md](integrations/audible.md) (scraping), [phase3/README.md](phase3/README.md) (automation)
 **"How do torrent downloads work?"** → [phase3/qbittorrent.md](phase3/qbittorrent.md), [backend/services/jobs.md](backend/services/jobs.md)

documentation/features/bookdate.md

@@ -390,6 +390,22 @@ Personalized audiobook discovery using OpenAI/Claude APIs. Admin configures AI p
   - Consistent with recommendations endpoint filtering behavior
 - Files updated: `src/app/api/bookdate/generate/route.ts:147-157`
 
+**8. Test Connection with Bad Custom LLM Credentials Logs User Out**
+- Issue: Testing custom LLM connection with invalid credentials causes user to be logged out instead of showing error
+- User Experience: "When I test connection with wrong API key, I get logged out of ReadMeABook instead of seeing an error message"
+- Cause: Custom LLM provider returns 401 Unauthorized for invalid credentials
+  - Test-connection endpoint passed through external service's 401 status code
+  - `fetchWithAuth()` utility intercepts ALL 401 responses, assuming they indicate expired user session
+  - Triggers automatic token refresh, then logout if refresh fails or still 401
+  - User logged out of application when only external service credentials were invalid
+- Fix: Return 400 Bad Request instead of passing through external service status codes
+  - Changed lines 196, 387 in `src/app/api/bookdate/test-connection/route.ts`
+  - Now returns `{ status: 400 }` for all custom provider connection failures
+  - Reserve 401 status exclusively for application authentication issues
+  - External service credential failures are client errors (400), not auth errors (401)
+  - Added tests to verify 401 from external provider returns 400 to client
+- Files updated: `src/app/api/bookdate/test-connection/route.ts:190-197,382-389`, `tests/api/bookdate-test-connection.routes.test.ts:254-294`
+
 ## Related
 
 - Full requirements: [features/bookdate-prd.md](bookdate-prd.md)

documentation/settings-pages.md

@@ -1,9 +1,67 @@
 # Settings Pages
 
-**Status:** ✅ Implemented
+**Status:** ✅ Implemented | ♻️ Refactored (Jan 2026)
 
 Single tabbed interface for admins to view/modify system configuration post-setup with mandatory validation before saving.
 
+## Architecture (Refactored Jan 2026)
+
+**Original:** Monolithic 2,971-line component
+**Current:** Modular architecture with 89% code reduction (2,971 → 325 lines)
+
+**Structure:**
+```
+src/app/admin/settings/
+├── page.tsx                      # Shell component (325 lines)
+├── lib/
+│   ├── types.ts                  # Shared TypeScript interfaces
+│   └── helpers.ts                # Business logic (206 lines)
+├── hooks/
+│   └── useSettings.ts            # Global settings hook
+└── tabs/                         # Feature modules
+    ├── LibraryTab/               # Plex/Audiobookshelf config
+    │   ├── LibraryTab.tsx
+    │   ├── useLibrarySettings.ts
+    │   ├── PlexSection.tsx
+    │   ├── AudiobookshelfSection.tsx
+    │   └── index.ts
+    ├── AuthTab/                  # Authentication (OIDC + Manual)
+    │   ├── AuthTab.tsx
+    │   ├── useAuthSettings.ts
+    │   ├── OIDCSection.tsx
+    │   ├── RegistrationSection.tsx
+    │   ├── PendingUsersTable.tsx
+    │   └── index.ts
+    ├── IndexersTab/              # Prowlarr/indexers
+    │   ├── IndexersTab.tsx
+    │   ├── useIndexersSettings.ts
+    │   └── index.ts
+    ├── DownloadTab/              # qBittorrent/SABnzbd
+    │   ├── DownloadTab.tsx
+    │   ├── useDownloadSettings.ts
+    │   └── index.ts
+    ├── PathsTab/                 # Directory paths
+    │   ├── PathsTab.tsx
+    │   ├── usePathsSettings.ts
+    │   └── index.ts
+    ├── EbookTab/                 # E-book sidecar
+    │   ├── EbookTab.tsx
+    │   ├── useEbookSettings.ts
+    │   └── index.ts
+    └── BookDateTab/              # AI recommendations
+        ├── BookDateTab.tsx
+        ├── useBookDateSettings.ts
+        └── index.ts
+```
+
+**Benefits:**
+- Single Responsibility: Each tab manages its own state/logic
+- Testability: Individual tabs can be unit tested
+- Maintainability: Changes to one feature don't affect others
+- Performance: Lazy loading possible (future optimization)
+- Reusability: Custom hooks can be used elsewhere
+- Code Quality: Follows React best practices
+
 ## Sections
 
 1. **Plex** - URL, token (masked), library ID, Audible region, filesystem scan trigger toggle

documentation/testing.md

@@ -0,0 +1,28 @@
+# Testing
+
+**Status:** ⏳ In Progress | Backend unit testing framework (Vitest)
+
+## Overview
+Unit tests for backend logic with isolated mocks (Prisma, integrations, queue).
+
+## Key Details
+- **Runner:** Vitest (`vitest.config.ts`, Node environment)
+- **Setup:** `tests/setup.ts` sets `NODE_ENV=test`, `TZ=UTC`, blocks unmocked fetch
+- **Helpers:** `tests/helpers/prisma.ts`, `tests/helpers/job-queue.ts`
+- **GitHub Actions:** Manual workflow `.github/workflows/manual-tests.yml` runs `npm test`
+- **Coverage:** `npm run test:coverage` (reports in `coverage/`)
+- **Scope:** Backend unit tests only; no real network or services
+
+## API/Interfaces
+```
+npm run test
+npm run test:watch
+npm run test:coverage
+```
+
+## Critical Issues
+- API route unit tests are incomplete; add route-level mocks before enforcing coverage.
+
+## Related
+- [backend/services/jobs.md](backend/services/jobs.md)
+- [backend/services/scheduler.md](backend/services/scheduler.md)