ReadMeABook/documentation/setup-wizard.md

Setup Wizard

Status: ✅ Implemented

9-step wizard for first-time configuration with connection testing, validation, database persistence, BookDate AI setup, and automated initial job execution.

Features

• 9 steps with progress indicator
• Connection testing for Plex, Prowlarr, qBittorrent
• Path validation with write permission checking
• Automated initial jobs (Audible refresh, Plex scan)
• Auto-enabling of scheduled jobs
• Dark mode support

Steps

1. Welcome - Intro screen
2. Admin Account - Create admin user
3. Plex (or Audiobookshelf) - Server URL, auth, library selection, filesystem scan trigger toggle
4. Prowlarr - URL, API key, indexer selection with priorities (1-25), seeding time, RSS monitoring
5. Download Client - qBittorrent/Transmission config
6. Paths - Download + media directories with validation
7. BookDate - AI-powered recommendations config (OpenAI/Claude, optional)
8. Review - Summary of all configs
9. Finalize - Run initial Audible refresh + Plex scan, enable scheduled jobs

API Endpoints

POST /api/setup/test-plex

• Tests Plex connection, returns libraries if successful

POST /api/setup/test-prowlarr

• Tests connection, returns indexer details (id, name, protocol)
• User selects indexers and assigns priorities

POST /api/setup/test-download-client

• Tests qBittorrent/Transmission, returns client version

POST /api/setup/complete

• Saves all config to database
• Creates admin user account
• Enables auto jobs (Plex scan, Audible refresh)
• Marks setup as complete
• Returns JWT tokens for auto-login

State Interface

interface SetupState {
  currentStep: number;
  plexUrl: string;
  plexToken: string;
  plexLibraryId: string;
  prowlarrUrl: string;
  prowlarrApiKey: string;
  prowlarrIndexers: Array<{id: number, name: string, priority: number, seedingTimeMinutes: number, rssEnabled: boolean}>;
  downloadClient: 'qbittorrent' | 'transmission';
  downloadClientUrl: string;
  downloadClientUsername: string;
  downloadClientPassword: string;
  downloadDir: string;
  mediaDir: string;
  validated: {plex: boolean, prowlarr: boolean, downloadClient: boolean, paths: boolean};
}

Validation

Plex: Valid URL, non-empty token, connection succeeds, library available Prowlarr: Valid URL, non-empty API key, connection succeeds, ≥1 indexer selected with priority 1-25, seedingTimeMinutes ≥0, rssEnabled boolean (RSS timing defaults to 15min, configurable in scheduled jobs) Download Client: Valid URL, credentials required, connection succeeds Paths: Absolute paths, writable

OIDC-Only Setup Flow

When using OIDC authentication without creating a local admin:

1. Setup wizard completes without creating admin user
2. FinalizeStep detects no access token (OIDC-only mode)
3. Shows message: jobs will run on first login
4. User completes setup → redirected to /login
5. User logs in via OIDC → first user becomes admin
6. Initial jobs (Audible refresh + Library scan) trigger automatically in background
7. User redirected to /setup/initializing page → shows real-time job progress
8. Jobs complete → user clicks "Go to Homepage" → fully initialized app

User Experience:

• FinalizeStep: Clear instructions about first login
• First OIDC login: Automatic redirect to initializing page
• Initializing page: Real-time job status with progress indicators
• Subsequent logins: Normal login flow (no initializing page)

Implementation:

• setup/page.tsx: Passes hasAdminTokens prop to FinalizeStep, clears localStorage to remove stale tokens
• FinalizeStep.tsx: Uses prop (not localStorage) to detect mode, shows appropriate UI
• OIDCAuthProvider.ts:
  • Triggers initial jobs on first user creation
  • Returns isFirstLogin: true flag in AuthResult
• api/auth/oidc/callback/route.ts:
  • Checks isFirstLogin flag
  • Redirects to /setup/initializing for first login
  • Normal redirect for subsequent logins
• setup/initializing/page.tsx:
  • Reads auth data from URL hash
  • Polls job status every 2s
  • Shows real-time progress
  • Auto-enables "Go to Homepage" when complete
• system.initial_jobs_run config flag prevents duplicate runs

Fixed Issues ✅

1. Plex Server Info Parsing

• Issue: "Connected to undefined undefined"
• Cause: XML parsing not extracting MediaContainer.$ attributes
• Fix: Proper XML attribute parsing with fallbacks

2. Auth Requirement

• Issue: Setup completion endpoint required auth before user login
• Fix: Removed auth requirement from /api/setup/complete

3. Plex Token Hint

• Issue: Incorrect path shown for finding token
• Fix: Link to official Plex documentation

4. Prowlarr Indexer Selection

• Feature: Added UI for selecting indexers with priorities (1-25)
• Auto-selects all with default priority 10
• Saves to database as JSON

5. Initial Job Execution

• Feature: Added FinalizeStep (step 9)
• Normal mode (with admin): Runs jobs during setup
• OIDC-only mode: Jobs run on first OIDC login
• Polls job status every 2s until actual completion
• Shows real-time execution status (pending → running → completed/failed)
• Prevents navigation until all jobs complete
• Uses /api/admin/job-status/:id endpoint for status polling

6. OIDC-Only Setup Support

• Issue: Initial jobs failed with "Authentication required" or "Failed to fetch job configuration"
• Root causes:
  • No admin user created during setup (OIDC-only), no auth token available
  • Stale tokens in localStorage from previous tests caused false-positive detection
• Fix: Proper architectural solution
  • Setup wizard passes hasAdminTokens prop to FinalizeStep (explicit mode detection)
  • Setup wizard clears localStorage before storing new tokens (removes stale data)
  • FinalizeStep uses prop instead of checking localStorage (avoids stale token issues)
  • OIDC-only mode redirects to /login after setup completion
• Jobs automatically trigger on first OIDC login (first user becomes admin)
• Background execution doesn't block authentication flow

7. Initializing Page Job Detection

• Issue: "Job did not start" error on initializing page while jobs running
• Root cause: lastRunJobId field missing from ScheduledJob schema
  • triggerJobNow() returned Bull job ID but never stored it
  • Initializing page couldn't find running jobs
• Fix: Database schema update + scheduler service update
  • Added lastRunJobId field to ScheduledJob model
  • Updated triggerJobNow() to store Bull job ID in database
  • Migration: 20251221072639_add_last_run_job_id_to_scheduled_jobs
• Initializing page now successfully finds and polls running jobs

Related Files

• /src/app/setup/ - Wizard components
• /src/app/setup/initializing/ - First login initialization page (OIDC-only)
• /src/app/api/setup/ - API routes
• /src/lib/services/auth/OIDCAuthProvider.ts - OIDC auth + first login detection
• /src/lib/services/auth/IAuthProvider.ts - Auth interfaces (includes isFirstLogin flag)