laiye-doc-processing

What This Skill Does

The laiye-doc-processing skill integrates the enterprise-grade Laiye Agentic Document Processing (ADP) engine directly into OpenClaw. It serves as a powerful API wrapper that transforms unstructured documents into structured, machine-readable data. By leveraging advanced Vision Language Models (VLM) and LLMs, the skill can process over 10 different file formats—including PDFs, images (JPEG, PNG, BMP), and office documents (DOCX, XLSX)—with high precision. Each extracted field comes with a confidence score, allowing developers to set automated verification thresholds for mission-critical business processes. Whether you are dealing with invoices, shipping manifests, or complex receipts, this skill eliminates manual data entry and configuration overhead.

Installation

You can install this skill via the OpenClaw command-line interface. Use the following command to integrate it into your local environment: clawhub install openclaw/skills/skills/jeane-li/laiye-doc-processing

Once installed, ensure you have your credentials ready. You will need an ADP_ACCESS_KEY, APP_KEY, and APP_SECRET. For security purposes, we highly recommend adding these as environment variables rather than hard-coding them into your scripts. Save them in your shell configuration profile, such as .bashrc or .zshrc, to ensure secure access across all your OpenClaw workflows.

Use Cases

This skill is designed for businesses looking to automate their document-heavy workflows:

• Accounts Payable Automation: Automatically extract invoice numbers, line items, tax details, and total amounts for ERP integration.
• Logistics & Supply Chain: Parse shipping orders and bills of lading to track inventory updates automatically.
• Data Digitization: Convert large batches of legacy scanned PDF archives into structured Excel or JSON datasets for analytical processing.
• Receipt Management: Extract expense information from travel and corporate receipts to streamline employee reimbursements.

Example Prompts

1. "Analyze this invoice at https://example.com/invoice123.pdf and give me the total amount and invoice date."
2. "Extract all line items from the provided receipt base64 data and store them in a JSON format."
3. "Process the attached contract and identify the renewal date, party names, and the primary jurisdiction field."

Tips & Limitations

• Security: Always rotate your APP_SECRET every 90 days. Keep your file permissions restricted (e.g., chmod 600) to prevent unauthorized access to your credential files.
• Performance: For high-volume processing, prefer passing file_url over raw file_base64 to minimize latency and memory overhead.
• Limitations: The accuracy of the extraction depends heavily on the image quality. While it performs exceptionally well on standard business documents, highly distorted or extremely low-resolution scans may result in lower confidence scores. Always implement logic in your agent to handle confidence < 0.8 cases by flagging them for human review.