extract

extract<T>(options: ExtractOptions): Promise<ExtractionRunResult<T>>

Extract structured data from the current page using a schema.

const result = await opensteer.extract({
  description: 'product-info',
  schema: {
    name: { type: 'string', description: 'Product name' },
    price: { type: 'string', description: 'Product price' },
  },
})
console.log(result.data) // { name: '...', price: '...' }

Array extraction:

const result = await opensteer.extract({
  description: 'search-results',
  schema: {
    items: [{
      title: { type: 'string' },
      url: { type: 'string' },
    }],
  },
})
console.log(result.data.items) // [{ title: '...', url: '...' }, ...]

Schema field options:

Option	Type	Description
type	'string' | 'number' | 'boolean'	Value type
description	string	Field description for LLM
element	number	Counter reference
selector	string	CSS selector
attribute	string	Element attribute to extract
source	'current_url'	Extract from page URL

extractFromPlan

extractFromPlan<T>(options: ExtractFromPlanOptions): Promise<ExtractionRunResult<T>>

Extract data using a pre-planned extraction (persisted schema and selectors).

const result = await opensteer.extractFromPlan({
  description: 'product-info',
  schema: { name: { type: 'string' }, price: { type: 'string' } },
})