Agent Skills: WorldCat API

WorldCat API

WorldCat is the world's largest library catalog with 500+ million bibliographic records and holdings from 10,000+ libraries worldwide.

Two Main APIs

Metadata API

For detailed bibliographic records and metadata.

Python Library: bookops-worldcat (recommended)

from bookops_worldcat import MetadataSession, WorldcatAccessToken

# Authenticate
token = WorldcatAccessToken(
    key=OCLC_CLIENT_ID,
    secret=OCLC_CLIENT_SECRET,
    scopes="WorldCatMetadataAPI"
)
session = MetadataSession(authorization=token)

# Search for books
response = session.brief_bibs_search(q="title:Moby Dick")

# Get classification
response = session.bib_get_classification(oclcNumber="123456")

# Check specific institutions
response = session.summary_holdings_search(
    oclcNumber="123456",
    heldBySymbol=["ZGM", "NYP"]
)

Key Methods:

• brief_bibs_search(q=query) - search bibliographic records
• summary_holdings_search() - check holdings at specific institutions
• bib_get_classification() - get LC/Dewey classification
• brief_bibs_get() - get single record by OCLC number

Documentation: https://bookops-cat.github.io/bookops-worldcat/

Search API V2

For comprehensive holdings data across ALL institutions.

Important: No Python library available - use direct REST calls.

Key Endpoint - Get ALL Holdings:

import requests
from bookops_worldcat import WorldcatAccessToken

# Authenticate
token = WorldcatAccessToken(
    key=OCLC_CLIENT_ID,
    secret=OCLC_CLIENT_SECRET,
    scopes='wcapi:view_holdings wcapi:view_institution_holdings'
)

# Get ALL institutions holding this item (ONE API call)
url = 'https://americas.discovery.api.oclc.org/worldcat/search/v2/bibs-holdings'
headers = {
    'Authorization': f'Bearer {token.token_str}',
    'Accept': 'application/json'
}
params = {'oclcNumber': '123456'}

response = requests.get(url, headers=headers, params=params)
data = response.json()

# Extract institution symbols
institutions = []
for record in data.get('briefRecords', []):
    for holding in record.get('institutionHolding', {}).get('briefHoldings', []):
        institutions.append({
            'symbol': holding.get('oclcSymbol'),
            'name': holding.get('institutionName'),
            'country': holding.get('country')
        })

symbols = [inst['symbol'] for inst in institutions]
# Returns: ["DLC", "NYP", "ZGM", "ZCU", ...]

Why This Matters: This endpoint returns ALL holdings in one API call, avoiding the need for 100+ individual requests when checking multiple institutions.

Authentication

All WorldCat APIs use OAuth 2.0 Client Credentials Grant:

# Set environment variables
export OCLC_CLIENT_ID="your_client_id"
export OCLC_CLIENT_SECRET="your_client_secret"

Get credentials from OCLC Developer Network: https://developer.api.oclc.org/

Common Patterns

Progressive Search Strategy

When searching with incomplete or messy citations:

1. Start broad - title keywords only
2. Add filters progressively if too many results
3. Pick best match based on metadata alignment

Example:

❌ Bad: "Ezra and Dorothy Pound: Letters in Captivity, 1945–1946"
✅ Good: "Pound Letters Captivity"
         → If 50+ results, add year: 1999
         → Pick match with correct publisher/author

Why: Bibliographic data has variations (punctuation, encoding, formatting). Better to get 20 results and filter than 0 results from being too specific.

Efficient Holdings Checking

Don't do this (slow - 150+ API calls):

for code in institution_codes:  # BAD!
    check_if_institution_holds(oclc_number, code)

Do this instead (fast - 1 API call):

# Get ALL holdings at once
holdings = get_all_holdings_via_search_api_v2(oclc_number)
symbols = set(holdings['institution_symbols'])

# Filter locally
tier_codes = {'NYP', 'ZCU', 'ZYU'}
matches = symbols & tier_codes

Handling Citations

Common citation formats encountered:

• (Year) Title, Publisher
• Title / Vol. N, Subtitle
• "Title" : Subtitle
• Author. (Year). Title.

Best practice: Extract clean search terms by removing volume info, subtitles, and complex punctuation initially. Add specificity only if needed.

Common Challenges

Encoding errors:

• Look for: ?, �, garbled characters
• Example: "Antigüedad" → "Antig?edad"
• Try: Remove accents initially, add back if no results

Multiple OCLC records:

• Same work may have multiple records (digital vs print editions)
• Check multiple if needed
• Document which record was used

Missing metadata:

• Not all records have DOIs or classification
• Handle missing data gracefully

Institution Symbols

Libraries have OCLC institution symbols (codes):

• DLC - Library of Congress
• NYP - New York Public Library
• ZGM - CUNY Graduate Center
• ZCU - Columbia University
• HUL - Harvard University

The Search API V2 /bibs-holdings endpoint returns these symbols. Users typically maintain lists of symbols for their consortia.

Workflows

Find and Identify Items

1. Search by ISBN if available (most reliable)
2. Otherwise keyword search with progressive narrowing
3. Extract OCLC number from best match

Get Complete Metadata

1. Use Metadata API or bookops-worldcat
2. Call brief_bibs_get(oclcNumber=num) or similar
3. Returns: title, author, publisher, ISBNs, subjects, classification

Check Holdings Across Institutions

1. Use Search API V2 /bibs-holdings endpoint
2. Returns complete array of institution symbols
3. Filter/group symbols locally as needed

Batch Processing

• Process items progressively
• Don't fail entire batch on one error
• Show progress for large datasets
• Document reasoning for each item

Reference

Official OCLC documentation:

• Metadata API: https://www.oclc.org/developer/api/oclc-apis/worldcat-metadata-api.en.html
• Search API V2: https://developer.api.oclc.org/wcv2