An asynchronous document extraction skill for AI agents to retrieve structured fields, bounding boxes, and specific key-value pairs from complex files.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install easydoc-extract
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install easydoc-extract using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is EasyLink EasyDoc Extract?
The EasyLink EasyDoc Extract skill is a production-grade integration designed for AI agents utilizing Openclaw Skills to seamlessly parse and structure document data. By leveraging the EasyLink async extraction API, this skill allows autonomous agents to ingest PDF documents, scanned images, business licenses, or organizational certificates, returning structured JSON outputs with highly reliable formatting.
Depending on the parameters provided, the skill automatically adapts to three distinct behaviors: closed extraction (using strict JSON schemas), prompt-driven extraction (via natural language prompts), or open extraction (auto-detecting all logical key-value pairs). This degree of flexibility makes it an invaluable utility for developer agents that need to turn messy, unstructured documentation into structured databases.
EasyLink EasyDoc Extract Use Cases
- Automated Invoice & Bill Processing: Extract billing names, dates, currencies, and total values from complex invoice files directly into standardized internal systems.
- KYC & Corporate Compliance Checks: Automatically parse official documents using dedicated structural modes like business license extraction (bl-extract) and organization-code certificate extraction (occ-extract).
- High-Precision Layout Processing: Capture visual metadata with bounding-box coordinate tracking using the specialized flash mode (easydoc-extract-flash).
- Schema-Enforced Document Audits: Feed arbitrary files to the extractor with a rigid JSON schema to isolate only specific fields required for database insertion.
How EasyLink EasyDoc Extract Works
- Input Validation: The skill checks for the necessary EASYLINK_API_KEY, validates that the file size is under the 100MB threshold, and checks the argument structure for the chosen mode.
- Async Task Submission: A multipart POST request is dispatched containing the files, the extraction mode, and any optional configurations (like a JSON schema or custom prompt).
- Asynchronous Polling: The skill queries GET /v1/easydoc/extract/{task_id} systematically, monitoring the status lifecycle until terminal completion (e.g., SUCCESS or FAILED).
- Format Normalization: Raw API payloads are sanitized and converted into a standard envelope containing page numbers and clean key-value dictionary fields.
EasyLink EasyDoc Extract Setup
Setup and Onboarding
If you do not have an active API key, register on the EasyLink platform:
- Open
https://platform.easylink-ai.comand register or log in. - Go to the API key management page and generate a new key.
- Export it as an environment variable in your terminal:
export EASYLINK_API_KEY="your_easylink_api_key_here"
Running Commands
You can interact with the API endpoints within your Openclaw Skills workflow using basic shell utilities:
# Trigger an open-extraction task
curl -X POST "https://api.easylink-ai.com/v1/easydoc/extract" \
-H "api-key: $EASYLINK_API_KEY" \
-F "[email protected]" \
-F "mode=easydoc-extract"
Alternatively, use the deterministic Python helper script to submit and poll in a single execution line:
# Closed extraction with custom field targets via Python
python3 scripts/easydoc_extract.py --file ./doc.pdf --mode easydoc-extract \
--fields "姓名" "日期" "金额" --save ./result.json
EasyLink EasyDoc Extract Data Schema & Taxonomy
Supported Extractor Modes
| Mode | Description | JSON Schema Requirement | Custom Prompts Support |
|---|---|---|---|
easydoc-extract |
Universal asynchronous extraction | Optional | Supported (when no schema provided) |
easydoc-extract-flash |
Bounding-box location extraction | Required | Not Supported |
bl-extract |
Fixed-field business license parser | Not Supported | Not Supported |
occ-extract |
Org-code certificate parser | Not Supported | Not Supported |
Output Schema Contract
The parsed output returns in a reliable format structured like the following schema:
{
"task_id": "string",
"status": "SUCCESS|ERROR|PENDING|PROCESSING|FAILED|COMPLETED|DONE",
"results": [
{
"page_number": 1,
"fields": {
"field_name": "value"
}
}
],
"raw": {}
}
EasyLink EasyDoc Extract Advanced Features
- Dynamic Schema Lock-down: Force target structures on unstructured documents by dynamically mapping fields to an exact custom JSON schema on demand.
- Self-Healing Polling Engine: Built-in state recovery logic monitors async job state changes, preventing workflow lockups and gracefully managing timeout scenarios.
- Normalized JSON Interfaces: Regardless of raw platform schema modifications, output contracts remain normalized to ease automated database ingestion.
- Multi-Agent Orchestration Support: Seamlessly plugs into Openclaw Skills multi-agent pipelines where document processing results feed downstream reasoning tasks.
SKILL.md
Loading
Related Openclaw Skills
EasyLink EasyDoc Mask
An automated AI agent skill to securely redact and mask sensitive personal or medical data in documents using the EasyLink asynchronous masking API.
Xiaohongshu Content Harvester & Scraper
A robust Xiaohongshu search, note reader, and user profile scraping skill integrated with the agent-browser JS environment.
Anti-Slop Frontend Skill
An opinionated frontend design skill designed to eliminate generic LLM aesthetic defaults and ship unique, production-grade landing pages, portfolios, and redesigns.
Dynamics 365 CRM
An advanced integration skill that empowers users to manage Microsoft Dynamics 365 CRM records, accounts, contacts, leads, and sales activities directly from their chat client.
Robinhood Agentic Trading (MCP)
Connects your AI agent directly to Robinhood's official agentic trading MCP server to securely read positions, analyze portfolios, and execute stock trades.
Xiaolongxia DeepSeek Programming Assistant
Xiaolongxia DeepSeek is an advanced, terminal-based AI programming assistant that integrates the DeepSeek API to automate coding, debugging, refactoring, and multi-turn developer interactions.
Featured*
