Skip to main content

Overview

adp-skills is a ready-to-use Skill collection from Laiye ADP. It combines vision language models, large language models, and agentic decision-making to upgrade rule-based field extraction into goal-driven document automation.

GitHub repository

laiye-ai-repos/adp-skill

Key advantages

Multimodal without labeling

Combine visual understanding and semantic understanding. Define extraction needs with natural language without sample labeling or model training.

100+ languages

Process Chinese, English, French, Southeast Asian, European, and mixed-language documents with one model.

Self-learning iteration

The agentic architecture records recognition deviations and improves with usage.

Ready to use

Start document processing tasks through conversational workflows.

Architecture: point, line, and surface

Skill categories

General workflow

Step 1: Install ADP CLI

Step 2: Get an API key

New users receive 100 free credits each month.

Step 3: Get an app ID

Connect to agent platforms

OpenClaw / Claude

Tencent WorkBuddy

Enter: I want to use this skill https://github.com/laiye-ai/adp-cli

Manus

Download the Skill package, upload it to the platform, and follow the prompts to configure authentication.

Internal enterprise platforms

Download the CLI Skill package from GitHub, NPM, or PyPI and add it to your agent platform.

Common scenarios

A finance user can say: “Summarize today’s Global Invoices into Excel.” ADP scans emails, finds attachments, extracts fields, cross-checks data, and generates a summary report.
A document specialist can say: “Check whether the shipper, receiver, and goods information are consistent across these documents.” ADP parses invoices, packing lists, delivery notes, and other documents, then outputs exceptions.

Resources

ClawHub

Install from ClawHub

GitHub Repo

View all Skill source code

Official website

ADP Global website

Feedback

GitHub Issues