跳至主要内容
小龙虾小龙虾AI
🤖

laiye-doc-processing

Enterprise-grade agentic document processing API. Accurately extracts key fields and line items from invoices, receipts, orders and more across 10+ file form...

下载97
星标4
版本1.2.0
自动化
安全通过
💬Prompt

技能说明


name: laiye-doc-processing description: Enterprise-grade agentic document processing API. Accurately extracts key fields and line items from invoices, receipts, orders and more across 10+ file formats, with confidence scoring. Zero-configuration, fast integration. Professionally optimized on massive enterprise documents. License: Commercial license required. New users receive 100 free credits monthly to offset usage.

Laiye Agentic Document Processing (ADP)

Agentic Document Processing API — convert 10+ file formats (.jpeg, .jpg, .png, .bmp, .tiff, .pdf, .doc, .docx, .xls, .xlsx) to structured JSON/Excel with per-field confidence scores using VLM and LLM.

Base URL: https://adp-global.laiye.com/?utm_source=clawhub

Quick Start

curl -X POST "https://adp-global.laiye.com/open/agentic_doc_processor/laiye/v1/app/doc/extract" \
  -H "Content-Type: application/json" \
  -H "X-Access-Key: $ADP_ACCESS_KEY" \
  -H "X-Timestamp: $(date +%s)" \
  -H "X-Signature: $(uuidgen)" \
  -d '{
    "app_key": "$ADP_APP_KEY",
    "app_secret": "$ADP_APP_SECRET",
    "file_url": "https://example.com/invoice.pdf"
  }'

Response:

{
  "status": "success",
  "extraction_result": [
    {
      "field_key": "invoice_number",
      "field_value": "INV-2024-001",
      "field_type": "text",
      "confidence": 0.95,
      "source_pages": [1]
    },
    {
      "field_key": "total_amount",
      "field_value": "1000.00",
      "field_type": "number",
      "confidence": 0.98,
      "source_pages": [1]
    }
  ]
}

Setup

1. Get Your API Credentials

# Contact ADP service provider to obtain:
# - app_key: Application access key
# - app_secret: Application secret key
# - X-Access-Key: Tenant-level access key

Save your credentials:

export ADP_ACCESS_KEY="your_access_key_here"
export ADP_APP_KEY="your_app_key_here"
export ADP_APP_SECRET="your_app_secret_here"

2. Configuration (Optional)

Recommended: Use environment variables (most secure):

{
  skills: {
    entries: {
      "adp-doc-extraction": {
        enabled: true,
        // API credentials loaded from environment variables
      },
    },
  },
}

Security Note:

  • Set file permissions: chmod 600 ~/.openclaw/openclaw.json
  • Never commit this file to version control
  • Prefer environment variables or secret stores
  • Rotate credentials regularly

Common Tasks

Extract from File URL

curl -X POST "https://adp-global.laiye.com/open/agentic_doc_processor/laiye/v1/app/doc/extract" \
  -H "Content-Type: application/json" \
  -H "X-Access-Key: $ADP_ACCESS_KEY" \
  -H "X-Timestamp: $(date +%s)" \
  -H "X-Signature: $(uuidgen)" \
  -d '{
    "app_key": "'"$ADP_APP_KEY"'",
    "app_secret": "'"$ADP_APP_SECRET"'",
    "file_url": "https://example.com/document.pdf"
  }'

Extract from Base64

# Convert file to base64
file_base64=$(base64 -i document.pdf | tr -d '\n')

curl -X POST "https://adp-global.laiye.com/open/agentic_doc_processor/laiye/v1/app/doc/extract" \
  -H "Content-Type: application/json" \
  -H "X-Access-Key: $ADP_ACCESS_KEY" \
  -H "X-Timestamp: $(date +%s)" \
  -H "X-Signature: $(uuidgen)" \
  -d "{
    \"app_key\": \"$ADP_APP_KEY\",
    \"app_secret\": \"$ADP_APP_SECRET\",
    \"file_base64\": \"$file_base64\",
    \"file_name\": \"document.pdf\"
  }"

Extract with VLM Results

curl -X POST "https://adp-global.laiye.com/open/agentic_doc_processor/laiye/v1/app/doc/extract" \
  -H "Content-Type: application/json" \
  -H "X-Access-Key: $ADP_ACCESS_KEY" \
  -H "X-Timestamp: $(date +%s)" \
  -H "X-Signature: $(uuidgen)" \
  -d '{
    "app_key": "'"$ADP_APP_KEY"'",
    "app_secret": "'"$ADP_APP_SECRET"'",
    "file_url": "https://example.com/document.pdf",
    "with_rec_result": true
  }'

Access VLM results: response["doc_recognize_result"]

Async Extraction (Large Documents)

Create extraction task:

curl -X POST "https://adp-global.laiye.com/open/agentic_doc_processor/laiye/v1/app/doc/extract/create/task" \
  -H "Content-Type: application/json" \
  -H "X-Access-Key: $ADP_ACCESS_KEY" \
  -H "X-Timestamp: $(date +%s)" \
  -H "X-Signature: $(uuidgen)" \
  -d '{
    "app_key": "'"$ADP_APP_KEY"'",
    "app_secret": "'"$ADP_APP_SECRET"'",
    "file_url": "https://example.com/large-document.pdf"
  }'

# Returns: {"task_id": "task_id_value", "metadata": {...}}

Poll for results:

curl -X GET "https://adp-global.laiye.com/open/agentic_doc_processor/laiye/v1/app/doc/extract/query/task/{task_id}" \
  -H "X-Access-Key: $ADP_ACCESS_KEY"

Advanced Features

Custom Scale Parameter

Enhance VLM quality with higher resolution:

# model_params: { "scale": 2.0 }

Specify Config Version

Use a specific extraction configuration:

# model_params: { "version_id": "config_version_id" }

Document Recognition Only

Get VLM results without extraction:

curl -X POST "https://adp-global.laiye.com/open/agentic_doc_processor/laiye/v1/app/doc/recognize" \
  -H "Content-Type: application/json" \
  -H "X-Access-Key: $ADP_ACCESS_KEY" \
  -H "X-Timestamp: $(date +%s)" \
  -H "X-Signature: $(uuidgen)" \
  -d '{
    "app_key": "'"$ADP_APP_KEY"'",
    "app_secret": "'"$ADP_APP_SECRET"'",
    "file_url": "https://example.com/document.pdf"
  }'

When to Use

Use ADP For:

  • Invoice processing
  • Order processing
  • Receipt processing
  • Financial document processing
  • Logistics document processing
  • Multi-table document data extraction

Don't Use For:

  • Video transcription
  • audio transcription

Best Practices

Document SizeEndpointNotes
Small files/doc/extract (sync)Immediate response
Large files/doc/extract/create/task (async)Poll for results

File Input:

  • file_url: Prefer for large files (already hosted)
  • file_base64: Use for direct upload (max 20MB)

Confidence Scores:

  • Range: 0-1 per field
  • Review fields with confidence <0.8 manually

Response Structure:

  • extraction_result: Array of extracted fields
  • doc_recognize_result: VLM results (when with_rec_result=true)
  • metadata: Processing info (pages, time, model)

Response Schema

Success Response

{
  "status": "success",
  "message": "string",
  "extraction_result": [
    {
      "field_key": "string",
      "field_value": "string",
      "field_type": "text|number|date|table",
      "confidence": 0.95,
      "source_pages": [1],
      "table_data": [...]  // for field_type="table"
    }
  ],
  "doc_recognize_result": [...],  // when with_rec_result=true
  "extract_config_version": "string",
  "metadata": {
    "total_pages": 5,
    "processing_time": 8.2,
    "model_used": "gpt-4o"
  }
}

Error Response

{
  "detail": "Error message description"
}

Common Use Cases

Invoice/Receipt Extraction

Extracts: invoice_number, invoice_date, vendor/customer_name, currency, vat_rate, total_amount_including_tax, total_amount_excluding_tax, line_items, etc.

Purchase Order Extraction

Extracts: order_number, order_date, buyer_name/seller_name, address, total_amount, line_items, etc.

Security & Privacy

Data Handling

Important: Documents uploaded to ADP are transmitted to https://adp-global.laiye.com/?utm_source=clawhub and processed on external servers.

Before uploading sensitive documents:

  • Review ADP privacy policy and data retention policies
  • Verify encryption in transit (HTTPS) and at rest
  • Confirm data deletion/retention timelines
  • Test with non-sensitive sample documents first

Best practices:

  • Do not upload highly sensitive PII until you've confirmed security posture
  • Use credentials with limited permissions if available
  • Rotate credentials regularly (every 90 days recommended)
  • Monitor API usage logs for unauthorized access
  • Never log or commit credentials to repositories

File Size Limits

  • Max file size: 50MB
  • Supported formats: .jpeg, .jpg, .png, .bmp, .tiff, .pdf, .doc, .docx, .xls, .xlsx
  • Concurrency limit: Free users support 1 concurrent request, paid users support 2 concurrent requests
  • Timeout: 10 minutes for sync requests

Operational Safeguards

  • Always use environment variables or secure secret stores for credentials
  • Never include real credentials in code examples or documentation
  • Use placeholder values like "your_access_key_here" in examples
  • Set appropriate file permissions on configuration files (600)
  • Enable credential rotation and monitor usage

Billing

Processing StageCost
Document Parsing0.5 credits/page
Purchase Order Extraction1.5 credits/page
Invoice/Receipt Extraction1.5 credits/page
Custom Extraction1 credit/page

New users: 100 free credits per month, no application restrictions.

Troubleshooting

Error CodeDescriptionCommon Causes & Solutions
400 Bad RequestInvalid request parameters• Missing app_key or app_secret<br>• Must provide exactly one input: file_url or file_base64<br>• Application has no published extraction config
401 UnauthorizedAuthentication failed• Invalid X-Access-Key<br>• Incorrect timestamp format (use Unix timestamp)<br>• Invalid signature format (must be UUID)
404 Not FoundResource not found• Application does not exist<br>• No published extraction config found for the application
500 Internal Server ErrorServer-side processing error• Document conversion failed<br>• VLM recognition timeout<br>• LLM extraction failure
Sync TimeoutRequest processing timed out• Large files should use async endpoint<br>• Poll /query/task/{task_id} for results

Pre-Publish Security Checklist

Before publishing or updating this skill, verify:

  • package.json declares requiredEnv and primaryEnv for credentials
  • package.json lists API endpoints in endpoints array
  • All code examples use placeholder values not real credentials
  • No credentials or secrets are embedded in SKILL.md or package.json
  • Security & Privacy section documents data handling and risks
  • Configuration examples include security warnings for plaintext storage
  • File permission guidance is included for config files

References

如何使用「laiye-doc-processing」?

  1. 打开小龙虾AI(Web 或 iOS App)
  2. 点击上方「立即使用」按钮,或在对话框中输入任务描述
  3. 小龙虾AI 会自动匹配并调用「laiye-doc-processing」技能完成任务
  4. 结果即时呈现,支持继续对话优化

相关技能