azure-ai-document-intelligence-ts
Extract structured data from documents with Azure AI
Manually processing invoices, receipts, and forms is slow and error-prone. This skill automates document analysis using Azure Document Intelligence to extract text, tables, and fields with high accuracy.
スキルZIPをダウンロード
Claudeでアップロード
設定 → 機能 → スキル → スキルをアップロードへ移動
オンにして利用開始
テストする
「azure-ai-document-intelligence-ts」を使用しています。 Analyze invoice from https://example.com/invoice-123.pdf
期待される結果:
Vendor: Contoso Electronics
Invoice Number: INV-2024-0892
Invoice Total: $1,247.50
Due Date: March 15, 2024
Confidence: 98.2%
「azure-ai-document-intelligence-ts」を使用しています。 Extract items from receipt image
期待される結果:
Merchant: Starbucks Store #4421
Date: Feb 20, 2024
Items:
- Caffee Latte (Large): $5.45
- Blueberry Muffin: $3.95
Subtotal: $9.40
Tax: $0.85
Total: $10.25
セキュリティ監査
安全All static analysis findings are false positives. The detected 'backtick execution' patterns are markdown code block delimiters, not shell commands. Network URLs point to legitimate Azure endpoints. Environment variable access follows standard credential management practices. No malicious patterns detected.
品質スコア
作れるもの
Accounts Payable Automation
Automatically extract vendor name, invoice total, and due dates from supplier invoices for accounting workflows
Expense Report Processing
Parse receipt images to extract merchant, items, and totals for expense management systems
Identity Verification
Extract name, date of birth, and ID numbers from driver licenses and passports for KYC workflows
これらのプロンプトを試す
Analyze this invoice PDF using Azure Document Intelligence. Extract the vendor name, invoice number, total amount, and due date. Return the results in a clear summary.
Process this receipt image and extract all line items with their descriptions and prices. Also extract the merchant name, transaction date, and total amount. List each item separately.
I have a folder of mixed documents (invoices, receipts, and contracts). Use a document classifier to identify each document type, then apply the appropriate prebuilt model to extract relevant fields from each. Summarize the results by document type.
Build a custom model to extract fields from our company's purchase order forms. The forms have fields for PO number, requester, department, line items with product codes and quantities, and approval signatures. Guide me through training data preparation and model deployment.
ベストプラクティス
- Always use getLongRunningPoller() for document analysis since operations are asynchronous
- Check confidence scores on extracted fields and set minimum thresholds for your use case
- Use prebuilt models when available before investing in custom model training
回避
- Do not skip error handling with isUnexpected() - always validate API responses
- Avoid polling with manual loops - use the built-in poller for long-running operations
- Do not hardcode credentials - use environment variables or Azure Key Vault for API keys