opensteer extract

Extract structured data from the page

Extracts structured data from the current page using a JSON schema definition.

Syntax

opensteer extract <schema> [options]

Arguments

• schema: JSON string defining the extraction schema

Options

• --session <id>: Session ID to use
• --description <text>: Description of what to extract
• --schema <json>: Alternative way to specify the schema

Examples

Simple Field Extraction

opensteer extract '{"title":"string","price":"number"}' --session agent-a

Product Information

opensteer extract '{
  "name": "string",
  "price": "number",
  "rating": "number",
  "inStock": "boolean"
}' --session agent-a

With Description

opensteer extract '{"email":"string","phone":"string"}' \
  --description "Contact information" \
  --session agent-a

Array Extraction

opensteer extract '{
  "products": [
    {
      "name": "string",
      "price": "number"
    }
  ]
}' --session agent-a

Using —schema Flag

opensteer extract --schema '{"title":"string"}' --session agent-a

Complex Nested Schema

opensteer extract '{
  "article": {
    "title": "string",
    "author": {
      "name": "string",
      "bio": "string"
    },
    "publishedDate": "string",
    "tags": ["string"]
  }
}' --session agent-a

Response

Returns the extracted data matching the schema:

{
  "ok": true,
  "data": {
    "title": "Example Product",
    "price": 29.99,
    "rating": 4.5,
    "inStock": true
  }
}

Schema Format

The schema is a JSON object that defines the structure of data to extract:

• Primitive types: "string", "number", "boolean"
• Objects: Nested objects with typed fields
• Arrays: [type] or [{object schema}]

Examples

// Simple fields
{
  "name": "string",
  "age": "number",
  "active": "boolean"
}

// Nested object
{
  "user": {
    "name": "string",
    "email": "string"
  }
}

// Array of strings
{
  "tags": ["string"]
}

// Array of objects
{
  "items": [
    {
      "id": "number",
      "name": "string"
    }
  ]
}

Use Cases

• Web scraping structured data from product pages
• Extracting article metadata and content
• Gathering contact information from directories
• Collecting pricing and availability data
• Building datasets from web pages