# Build Azure Document Intelligence Workflows

Document extraction workflows require precise Azure REST SDK patterns and asynchronous operation handling. This skill provides practical TypeScript guidance for analysis, classification, and custom models.

## Install

```bash
npx skillstore add sickn33/azure-ai-document-intelligence-ts
```

## Metadata

- Status: approved
- Slug: sickn33-azure-ai-document-intelligence-ts
- Skillstore revision: r2
- Version status: missing
- Tree hash: eb5318fdefb46e75a95afbb03bdd8a46458b5b58857eca04cc05c1eaa205501e
- Author: sickn33
- GitHub username: sickn33
- License: MIT
- Repository: https://github.com/sickn33/antigravity-awesome-skills/tree/main/skills/azure-ai-document-intelligence-ts
- Ref: f9e2c34b4f19c7f3e6b0a1e93227b5f77cc12526
- Supported tools: Claude, Codex, Claude Code
- Audit status: complete
- Agent install advisory: allowed
- Manual install advisory: allowed
- Artifact signature: available
- Audit attestation: unavailable
- Human verification: not\_verified
- Risk factors: external\_commands, network, env\_access
- Quality score: 78
- Quality tier: bronze
- Public page: https://skillstore.pages.dev/skills/sickn33-azure-ai-document-intelligence-ts
- Manifest: https://skillstore.pages.dev/api/skills/sickn33-azure-ai-document-intelligence-ts/manifest

## Capabilities

- Configures the Azure Document Intelligence REST client with DefaultAzureCredential or an API key.
- Analyzes documents from remote URLs or local files with prebuilt Azure models.
- Extracts page, table, invoice, receipt, and line-item information from analysis results.
- Lists available document models and reads custom model service limits.
- Builds custom models and document classifiers from Azure Blob Storage training data.
- Uses Azure long-running pollers and typed error checks for asynchronous operations.

## Use Cases

- Automate Invoice Extraction: Create a TypeScript workflow that extracts vendor, total, due date, and line-item fields from invoices.
- Classify Incoming Documents: Build and use a classifier that separates invoices, receipts, and other document types stored in Azure.
- Train Specialized Models: Configure custom template or neural models from labeled documents in Azure Blob Storage.

## Prompt Templates

### Configure the Client

```
Show me how to install and configure the Azure Document Intelligence REST client in TypeScript using DefaultAzureCredential and an environment-based endpoint.
```

### Analyze a Local Invoice

```
Create a TypeScript example that reads [file path], analyzes it with prebuilt-invoice, checks errors, polls completion, and reports key invoice fields.
```

### Build a Document Classifier

```
Design a classifier for [document types] using training data under [Azure Blob prefixes]. Include environment configuration, polling, and result checks.
```

### Design a Production Analysis Module

```
Design a typed TypeScript module for URL and local-file analysis using [model ID]. Include authentication choices, error handling, polling, pagination, and confidence thresholds.
```

## Limitations

- Requires an Azure Document Intelligence resource, valid credentials, and the listed npm packages.
- Examples are implementation patterns, not a complete application with validation, retries, monitoring, or deployment configuration.
- Local file analysis sends encoded document content to Azure, so users must apply privacy, residency, and retention controls.
- Extraction quality varies by document and model; confidence thresholds and human review remain application responsibilities.

## Best Practices

- Prefer DefaultAzureCredential when managed identity or developer identity is available.
- Check isUnexpected before creating a poller, and handle failed long-running operations explicitly.
- Select the narrowest suitable model and enforce field confidence thresholds for the business workflow.

## Anti Patterns

- Do not hardcode API keys, SAS URLs, or resource-specific endpoints in source files.
- Do not send regulated or confidential documents without approved Azure privacy and residency controls.
- Do not assume extracted fields are correct when confidence values or required fields are missing.

## Security Audit

- Audited at: 2026-07-23T19:19:00.092\+00:00
- Summary: All 59 static alerts are false positives caused by Markdown syntax, placeholders, or standard environment-based configuration. The examples use official Azure packages and expose no credentials. No prompt injection, exfiltration intent, or other semantic threat was found.

## Stats

- Views: 157
- Downloads: 8
- Favorites: 0
- Popularity score: 0