What is the Cataloging Assistant?
- AI-Powered LCSH Generation — Uses AI models to analyze bibliographic information and suggest appropriate Library of Congress Subject Headings
- LOC Validation — Validates every suggestion against the Library of Congress authorities database (LCSH & LCNAF)
- MARC Record Generation — Automatically generates properly formatted MARC records (650/600/610/651 tags)
- Privacy-First Design — All data stored locally in your browser. Zero-knowledge architecture
- Progressive Web App — Install on your device and use like a native app
App Navigation
Home
Landing page with quick-access cards to start a new session, view history, or open settings.
Wizard
The core 3-step workflow: enter bibliographic data, review AI suggestions, and export MARC records.
History
Review all past cataloging sessions. View details, export to CSV, or delete old sessions.
Settings
Configure your AI model provider, manage API keys, and customize the system prompt.
Tip: The active page is highlighted with a dark background in the navigation bar. You can navigate between pages at any time without losing your wizard progress.
Settings: Select a Provider
Available Providers
- OpenAI — GPT-4o, GPT-4, GPT-3.5
- Google Gemini — Gemini 2.5 Flash/Pro, 2.0 Flash
- DeepSeek — DeepSeek Chat, Reasoner
- Qwen — Qwen models (Alibaba)
- Z.ai, Kimi, Minimax — Additional providers
How to: Click the "Provider" dropdown, then select your preferred provider from the list.
Settings: Add Your API Key
Adding a Key
- Enter your API key in the "Enter API key" field
- Optionally add a label (e.g., "Work Key", "Personal Key")
- Click "Add" to save the key
- Keys are masked in the UI for security (XXXX...YYYY)
- The first key added is automatically set as Default
Security: Keys are stored only in your browser's local storage. They never leave your device except when making requests to the AI provider.
The 3-Step Wizard Workflow
1
Bibliographic Information
Enter title, author, abstract, table of contents, notes. Upload images.
➔
2
Validated Suggestions
Review AI suggestions validated against LOC with similarity scores.
➔
3
Final Recommendations
MARC records, copy to clipboard, export CSV, save to history.
A visual progress bar tracks your progress. You can navigate back to previous steps using the "Back" button at any time.
Step 1: Enter Bibliographic Information
Input Fields
| Field | Required? |
|---|---|
| Title | Yes (or upload image) |
| Author | No |
| Abstract | No (recommended) |
| Table of Contents | No |
| Additional Notes | No |
| Images | No (PNG/JPEG) |
Tip: The more detail you provide (especially abstract & table of contents), the better the AI suggestions.
Image Upload Note: Image analysis only works with AI models that have vision capabilities (e.g., GPT-4o, Gemini 2.5 Flash/Pro). Text-only models will ignore uploaded images.
Step 1: Generating Suggestions
What Happens Behind the Scenes
1
Your bibliographic data is sent to the configured AI model
2
AI analyzes the content and generates subject heading suggestions
3
Each suggestion is validated against the LOC authorities API (LCSH & LCNAF)
4
Similarity scores are calculated using Levenshtein distance
5
Results auto-advance to Step 2
Processing typically takes 10-30 seconds depending on the AI provider and input complexity.
Step 2: Validated Suggestions
Results for "The Great Gatsby"
- 96% Excellent Match — 8 of 8 terms validated
- Subject Analysis — AI-generated summary of the work's themes
- Each term shows: heading name, LCSH source badge, and similarity %
- AI Reasoning explains why each heading was chosen
- "View on LOC" links directly to id.loc.gov authority records
Green checkmarks indicate exact LOC matches. The AI Additional badge marks terms the AI suggested beyond the input.
Step 2: Individual Term Details
Terms with 100% matches and AI reasoning
"AI Additional" terms and "Generate MARC Records" button
Click "Generate MARC Records" at the bottom to proceed to Step 3 and create MARC records for all validated terms.
Step 3: Final Recommendations
- Subject Analysis — AI expert analysis of the work's themes
- Overall Validation Score — 96% Excellent Match
- Each term shows its MARC record with correct tag formatting
- MARC tags: 650 for LCSH subjects, 600/610 for LCNAF names
- Individual "Copy" button for each MARC record
Example MARC:
650 _0 $a American fiction $y 20th century
Step 3: Export Options
Copy All MARC
Copy all MARC records to your clipboard at once for pasting into your ILS.
Export CSV
Download a CSV file with all terms, matches, sources, reasoning, and MARC records.
Save & View History
Persist the entire session for future reference, review, and re-export.
Individual Copy
Copy a single MARC record with the "Copy" button next to each term.
The "Back" button lets you return to Step 2 to review suggestions before exporting.
Conversation History
History Table
- View all past sessions with date, title, author, and term count
- Click the eye icon to view full session details
- Click the download icon to export a session as CSV
- Click the trash icon to delete (with confirmation)
Bulk Actions
- "Export All" — Download all sessions as a single CSV
- "Clear All" — Delete entire history (with confirmation)
History: Session Detail View
Detail Dialog Shows
- Title & timestamp of the session
- "Export CSV" and "Copy All MARC" buttons at the top
- Bibliographic Information — title, author, and all input data
- Subject Analysis — AI-generated thematic analysis
- Recommended Terms — each with source badge, similarity score, LOC link, AI reasoning, and MARC record
- Individual copy buttons for each MARC record
Tip: Export important sessions to CSV regularly as a backup. Browser data clearing will erase local history.
