Description

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

README (SKILL.md)

\r \r

Laiye Agentic Document Processing (ADP)\r

Name: Adp Skill
Author: anna4lucky-stack

\r 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.\r \r

Base URL: https://adp-global.laiye.com/?utm_source=github\r \r

Quick Start\r

\r

curl -X POST "https://adp-global.laiye.com/open/agentic_doc_processor/laiye/v1/app/doc/extract" \\r
  -H "Content-Type: application/json" \\r
  -H "X-Access-Key: $ADP_ACCESS_KEY" \\r
  -H "X-Timestamp: $(date +%s)" \\r
  -H "X-Signature: $(uuidgen)" \\r
  -d '{\r
    "app_key": "$ADP_APP_KEY",\r
    "app_secret": "$ADP_APP_SECRET",\r
    "file_url": "https://example.com/invoice.pdf"\r
  }'\r
```\r
\r
Response:\r
```json\r
{\r
  "status": "success",\r
  "extraction_result": [\r
    {\r
      "field_key": "invoice_number",\r
      "field_value": "INV-2024-001",\r
      "field_type": "text",\r
      "confidence": 0.95,\r
      "source_pages": [1]\r
    },\r
    {\r
      "field_key": "total_amount",\r
      "field_value": "1000.00",\r
      "field_type": "number",\r
      "confidence": 0.98,\r
      "source_pages": [1]\r
    }\r
  ]\r
}\r
```\r
\r
## Setup\r
\r
### 1. Get Your API Credentials\r
\r
```bash\r
# Contact ADP service provider to obtain:\r
# - app_key: Application access key\r
# - app_secret: Application secret key\r
# - X-Access-Key: Tenant-level access key\r
```\r
\r
Save your credentials:\r
```bash\r
export ADP_ACCESS_KEY="your_access_key_here"\r
export ADP_APP_KEY="your_app_key_here"\r
export ADP_APP_SECRET="your_app_secret_here"\r
```\r
\r
### 2. Configuration (Optional)\r
\r
**Recommended: Use environment variables** (most secure):\r
```json5\r
{\r
  skills: {\r
    entries: {\r
      "adp-doc-extraction": {\r
        enabled: true,\r
        // API credentials loaded from environment variables\r
      },\r
    },\r
  },\r
}\r
```\r
\r
**Security Note:**\r
- Set file permissions: `chmod 600 ~/.openclaw/openclaw.json`\r
- Never commit this file to version control\r
- Prefer environment variables or secret stores\r
- Rotate credentials regularly\r
\r
## Common Tasks\r
\r
### Extract from File URL\r
\r
```bash\r
curl -X POST "https://adp-global.laiye.com/open/agentic_doc_processor/laiye/v1/app/doc/extract" \\r
  -H "Content-Type: application/json" \\r
  -H "X-Access-Key: $ADP_ACCESS_KEY" \\r
  -H "X-Timestamp: $(date +%s)" \\r
  -H "X-Signature: $(uuidgen)" \\r
  -d '{\r
    "app_key": "'"$ADP_APP_KEY"'",\r
    "app_secret": "'"$ADP_APP_SECRET"'",\r
    "file_url": "https://example.com/document.pdf"\r
  }'\r
```\r
\r
### Extract from Base64\r
\r
```bash\r
# Convert file to base64\r
file_base64=$(base64 -i document.pdf | tr -d '\
')\r
\r
curl -X POST "https://adp-global.laiye.com/open/agentic_doc_processor/laiye/v1/app/doc/extract" \\r
  -H "Content-Type: application/json" \\r
  -H "X-Access-Key: $ADP_ACCESS_KEY" \\r
  -H "X-Timestamp: $(date +%s)" \\r
  -H "X-Signature: $(uuidgen)" \\r
  -d "{\r
    \"app_key\": \"$ADP_APP_KEY\",\r
    \"app_secret\": \"$ADP_APP_SECRET\",\r
    \"file_base64\": \"$file_base64\",\r
    \"file_name\": \"document.pdf\"\r
  }"\r
```\r
\r
### Extract with VLM Results\r
\r
```bash\r
curl -X POST "https://adp-global.laiye.com/open/agentic_doc_processor/laiye/v1/app/doc/extract" \\r
  -H "Content-Type: application/json" \\r
  -H "X-Access-Key: $ADP_ACCESS_KEY" \\r
  -H "X-Timestamp: $(date +%s)" \\r
  -H "X-Signature: $(uuidgen)" \\r
  -d '{\r
    "app_key": "'"$ADP_APP_KEY"'",\r
    "app_secret": "'"$ADP_APP_SECRET"'",\r
    "file_url": "https://example.com/document.pdf",\r
    "with_rec_result": true\r
  }'\r
```\r
\r
Access VLM results: `response["doc_recognize_result"]`\r
\r
### Async Extraction (Large Documents)\r
\r
**Create extraction task:**\r
```bash\r
curl -X POST "https://adp-global.laiye.com/open/agentic_doc_processor/laiye/v1/app/doc/extract/create/task" \\r
  -H "Content-Type: application/json" \\r
  -H "X-Access-Key: $ADP_ACCESS_KEY" \\r
  -H "X-Timestamp: $(date +%s)" \\r
  -H "X-Signature: $(uuidgen)" \\r
  -d '{\r
    "app_key": "'"$ADP_APP_KEY"'",\r
    "app_secret": "'"$ADP_APP_SECRET"'",\r
    "file_url": "https://example.com/large-document.pdf"\r
  }'\r
\r
# Returns: {"task_id": "task_id_value", "metadata": {...}}\r
```\r
\r
**Poll for results:**\r
```bash\r
curl -X GET "https://adp-global.laiye.com/open/agentic_doc_processor/laiye/v1/app/doc/extract/query/task/{task_id}" \\r
  -H "X-Access-Key: $ADP_ACCESS_KEY"\r
```\r
\r
## Advanced Features\r
\r
### Custom Scale Parameter\r
\r
Enhance VLM quality with higher resolution:\r
```bash\r
# model_params: { "scale": 2.0 }\r
```\r
\r
### Specify Config Version\r
\r
Use a specific extraction configuration:\r
```bash\r
# model_params: { "version_id": "config_version_id" }\r
```\r
\r
### Document Recognition Only\r
\r
Get VLM results without extraction:\r
```bash\r
curl -X POST "https://adp-global.laiye.com/open/agentic_doc_processor/laiye/v1/app/doc/recognize" \\r
  -H "Content-Type: application/json" \\r
  -H "X-Access-Key: $ADP_ACCESS_KEY" \\r
  -H "X-Timestamp: $(date +%s)" \\r
  -H "X-Signature: $(uuidgen)" \\r
  -d '{\r
    "app_key": "'"$ADP_APP_KEY"'",\r
    "app_secret": "'"$ADP_APP_SECRET"'",\r
    "file_url": "https://example.com/document.pdf"\r
  }'\r
```\r
\r
## When to Use\r
\r
### Use ADP For:\r
- Invoice processing\r
- Order processing\r
- Receipt processing\r
- Financial document processing\r
- Logistics document processing\r
- Multi-table document data extraction\r
\r
### Don't Use For:\r
- Video transcription\r
- audio transcription\r
\r
## Best Practices\r
\r
| Document Size | Endpoint | Notes |\r
|---------------|----------|-------|\r
| Small files | `/doc/extract` (sync) | Immediate response |\r
| Large files | `/doc/extract/create/task` (async) | Poll for results |\r
\r
**File Input:**\r
- `file_url`: Prefer for large files (already hosted)\r
- `file_base64`: Use for direct upload (max 20MB)\r
\r
**Confidence Scores:**\r
- Range: 0-1 per field\r
- Review fields with confidence \x3C0.8 manually\r
\r
**Response Structure:**\r
- `extraction_result`: Array of extracted fields\r
- `doc_recognize_result`: VLM results (when `with_rec_result=true`)\r
- `metadata`: Processing info (pages, time, model)\r
\r
## Response Schema\r
\r
### Success Response\r
```json\r
{\r
  "status": "success",\r
  "message": "string",\r
  "extraction_result": [\r
    {\r
      "field_key": "string",\r
      "field_value": "string",\r
      "field_type": "text|number|date|table",\r
      "confidence": 0.95,\r
      "source_pages": [1],\r
      "table_data": [...]  // for field_type="table"\r
    }\r
  ],\r
  "doc_recognize_result": [...],  // when with_rec_result=true\r
  "extract_config_version": "string",\r
  "metadata": {\r
    "total_pages": 5,\r
    "processing_time": 8.2,\r
    "model_used": "gpt-4o"\r
  }\r
}\r
```\r
\r
### Error Response\r
```json\r
{\r
  "detail": "Error message description"\r
}\r
```\r
\r
## Common Use Cases\r
\r
### Invoice/Receipt Extraction\r
Extracts: invoice_number, invoice_date, vendor/customer_name, currency, vat_rate, total_amount_including_tax, total_amount_excluding_tax, line_items, etc.\r
\r
### Purchase Order Extraction\r
Extracts: order_number, order_date, buyer_name/seller_name, address, total_amount, line_items, etc.\r
\r
## Security & Privacy\r
\r
### Data Handling\r
\r
**Important:** Documents uploaded to ADP are transmitted to `https://adp-global.laiye.com/?utm_source=github` and processed on external servers.\r
\r
**Before uploading sensitive documents:**\r
- Review ADP privacy policy and data retention policies\r
- Verify encryption in transit (HTTPS) and at rest\r
- Confirm data deletion/retention timelines\r
- Test with non-sensitive sample documents first\r
\r
**Best practices:**\r
- Do not upload highly sensitive PII until you've confirmed security posture\r
- Use credentials with limited permissions if available\r
- Rotate credentials regularly (every 90 days recommended)\r
- Monitor API usage logs for unauthorized access\r
- Never log or commit credentials to repositories\r
\r
### File Size Limits\r
\r
- **Max file size:** 50MB\r
- **Supported formats:** .jpeg, .jpg, .png, .bmp, .tiff, .pdf, .doc, .docx, .xls, .xlsx\r
- **Concurrency limit:** Free users support 1 concurrent request, paid users support 2 concurrent requests\r
- **Timeout:** 10 minutes for sync requests\r
\r
### Operational Safeguards\r
\r
- Always use environment variables or secure secret stores for credentials\r
- Never include real credentials in code examples or documentation\r
- Use placeholder values like `"your_access_key_here"` in examples\r
- Set appropriate file permissions on configuration files (600)\r
- Enable credential rotation and monitor usage\r
\r
## Billing\r
\r
| Processing Stage | Cost |\r
|-----------------|------|\r
| Document Parsing | 0.5 credits/page |\r
| Purchase Order Extraction | 1.5 credits/page |\r
| Invoice/Receipt Extraction | 1.5 credits/page |\r
| Custom Extraction | 1 credit/page |\r
\r
**New users:** 100 free credits per month, no application restrictions.\r
\r
## Troubleshooting\r
\r
| Error Code | Description | Common Causes & Solutions |\r
|------------|-------------|---------------------------|\r
| **400 Bad Request** | Invalid request parameters | • Missing `app_key` or `app_secret`\x3Cbr>• Must provide exactly one input: `file_url` or `file_base64`\x3Cbr>• Application has no published extraction config |\r
| **401 Unauthorized** | Authentication failed | • Invalid `X-Access-Key`\x3Cbr>• Incorrect timestamp format (use Unix timestamp)\x3Cbr>• Invalid signature format (must be UUID) |\r
| **404 Not Found** | Resource not found | • Application does not exist\x3Cbr>• No published extraction config found for the application |\r
| **500 Internal Server Error** | Server-side processing error | • Document conversion failed\x3Cbr>• VLM recognition timeout\x3Cbr>• LLM extraction failure |\r
| **Sync Timeout** | Request processing timed out | • Large files should use async endpoint\x3Cbr>• Poll `/query/task/{task_id}` for results |\r
\r
## Pre-Publish Security Checklist\r
\r
Before publishing or updating this skill, verify:\r
\r
- [ ] `package.json` declares `requiredEnv` and `primaryEnv` for credentials\r
- [ ] `package.json` lists API endpoints in `endpoints` array\r
- [ ] All code examples use placeholder values not real credentials\r
- [ ] No credentials or secrets are embedded in `SKILL.md` or `package.json`\r
- [ ] Security & Privacy section documents data handling and risks\r
- [ ] Configuration examples include security warnings for plaintext storage\r
- [ ] File permission guidance is included for config files\r
\r
## References\r
\r
- **API Base URL:** https://adp-global.laiye.com/\r
- **Flow Documentation:** [ADP Flow Index](../../refers/backend/flows/ADP-Flow-Index.md)\r
- **Extraction Flow:** [Document Extraction Flow](../../refers/backend/flows/09-Doc-Extraction-Flow.md)\r
- **Recognition Flow:** [Document Recognition Flow](../../refers/backend/flows/08-Doc-Recognition-Flow.md)\r

Usage Guidance

This skill appears to be a wrapper for an external Laiye ADP API and legitimately needs three API credentials, but the registry metadata does not declare those env vars and the package origin is not clearly verified. Before installing: (1) verify the skill source (official repo or vendor) and confirm the domain (adp-global.laiye.com) is correct and authorized by the vendor; (2) create scoped, revocable API keys for testing (do not reuse higher-privilege keys); (3) avoid sending sensitive or regulated documents until you confirm the vendor's data-retention and privacy policies; (4) prefer least-privilege credentials and enable rotation/monitoring; (5) if you cannot validate the author/hosting, do not install in production. If you decide to proceed, test with non-sensitive sample documents first and monitor network activity and API usage.

Capability Analysis

Type: OpenClaw Skill Name: adp-skill Version: 1.0.0 The skill bundle provides documentation and integration instructions for the Laiye Agentic Document Processing (ADP) API. It contains standard API usage examples using curl, requires user-provided credentials via environment variables, and includes explicit security warnings regarding data privacy and credential management in SKILL.md and README.md. No malicious code, obfuscation, or prompt injection attempts were identified.

Capability Assessment

⚠ Purpose & Capability

The skill's name/description and SKILL.md consistently describe an external document-extraction API that legitimately needs ADP credentials. However, the registry metadata reported at the top of the submission declares no required environment variables or primary credential, which conflicts with SKILL.md/README that require ADP_ACCESS_KEY, ADP_APP_KEY, and ADP_APP_SECRET. The mismatch between declared registry requirements and the actual instructions is an incoherence.

✓ Instruction Scope

SKILL.md is instruction-only and instructs the agent to POST document content (file URL or base64) to an explicit external API endpoint (https://adp-global.laiye.com). All referenced data (files, credentials) are consistent with the stated purpose of document extraction; there are no instructions to read unrelated system files or to exfiltrate unrelated secrets.

✓ Install Mechanism

No install spec or code is written to disk (instruction-only skill). This minimizes install-time risk because nothing is downloaded or executed on install.

⚠ Credentials

The SKILL.md and READMEs require three credentials (ADP_ACCESS_KEY, ADP_APP_KEY, ADP_APP_SECRET) which are proportional to calling the described external API. However, registry metadata incorrectly lists no required env vars — that discrepancy is concerning because it can mislead permission reviews and automated gating. Also the skill will send document data to an external domain, so credential scope and data-sensitivity controls matter.

✓ Persistence & Privilege

The skill does not request always:true and is user-invocable with normal autonomous invocation defaults. It does not ask to modify other skills or system configs. No elevated persistence privileges are requested.

Version History

v1.0.0

Initial release of the Laiye Agentic Document Processing (ADP) skill. - Provides enterprise-grade, agentic document processing API for extracting structured data (fields and line items) from various document types. - Supports 10+ file formats (.jpeg, .pdf, .docx, .xlsx, etc.) and outputs JSON/Excel with per-field confidence scores. - Includes quick integration guides for extracting data via file URL or base64 upload, with options for synchronous and asynchronous processing. - Documents API usage for invoices, receipts, orders, financial and logistics documents; not designed for video or audio transcription. - Features best practices for credential management, security, error handling, and production use cases. - 100 free credits monthly for new users; commercial license required for continued use.

Metadata

Slug adp-skill

Version 1.0.0

License MIT-0

All-time Installs 0

Active Installs 0

Total Versions 1

Frequently Asked Questions

What is Adp Skill?

Enterprise-grade agentic document processing API. Accurately extracts key fields and line items from invoices, receipts, orders and more across 10+ file form... It is an AI Agent Skill for Claude Code / OpenClaw, with 208 downloads so far.

How do I install Adp Skill?

Run "/install adp-skill" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.

Is Adp Skill free?

Yes, Adp Skill is completely free, licensed under MIT-0. You can download, install and use it at no cost.

Which platforms does Adp Skill support?

Adp Skill is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).

Who created Adp Skill?

It is built and maintained by anna4lucky-stack (@anna4lucky-stack); the current version is v1.0.0.

More Skills

Adp Skill