Page Querying

Fresh Source: docs.airtop.ai

Use pageQuery to ask natural language questions about page content. Use paginatedExtraction for multi-page data collection. Both support JSON schema output for structured results.

pageQuery — Single Page

Basic Query

TypeScript

const result = await client.windows.pageQuery(sessionId, windowId, {
  prompt: 'What is the main heading on this page?',
});
console.log(result.data.modelResponse);

Python

result = client.windows.page_query(
    session_id, window_id,
    prompt="What is the main heading on this page?"
)
print(result.data.model_response)

Structured Output with JSON Schema

TypeScript

const result = await client.windows.pageQuery(sessionId, windowId, {
  prompt: 'Extract all team members with their name, role, and LinkedIn URL',
  configuration: {
    outputSchema: {
      type: 'object',
      properties: {
        teamMembers: {
          type: 'array',
          items: {
            type: 'object',
            properties: {
              name: { type: 'string' },
              role: { type: 'string' },
              linkedinUrl: { type: 'string' },
            },
            required: ['name', 'role'],
          },
        },
      },
    },
  },
});
const data = JSON.parse(result.data.modelResponse);
console.log(data.teamMembers);

Python

result = client.windows.page_query(
    session_id, window_id,
    prompt="Extract all team members with their name, role, and LinkedIn URL",
    configuration={
        "outputSchema": {
            "type": "object",
            "properties": {
                "teamMembers": {
                    "type": "array",
                    "items": {
                        "type": "object",
                        "properties": {
                            "name": {"type": "string"},
                            "role": {"type": "string"},
                            "linkedinUrl": {"type": "string"}
                        },
                        "required": ["name", "role"]
                    }
                }
            }
        }
    }
)
import json
data = json.loads(result.data.model_response)
print(data["teamMembers"])

paginatedExtraction — Multi-Page

Automatically navigates through paginated content, collecting data from each page.

TypeScript

const result = await client.windows.paginatedExtraction(
  sessionId,
  windowId,
  {
    config: {
      outputSchema: {
        type: 'object',
        properties: {
          listings: {
            type: 'array',
            items: {
              type: 'object',
              properties: {
                title: { type: 'string' },
                price: { type: 'string' },
                location: { type: 'string' },
              },
            },
          },
        },
      },
    },
  }
);
console.log('Total items collected:', result.data.listings?.length);

Python

result = client.windows.paginated_extraction(
    session_id, window_id,
    config={
        "outputSchema": {
            "type": "object",
            "properties": {
                "listings": {
                    "type": "array",
                    "items": {
                        "type": "object",
                        "properties": {
                            "title": {"type": "string"},
                            "price": {"type": "string"},
                            "location": {"type": "string"}
                        }
                    }
                }
            }
        }
    }
)

Writing Effective Prompts

Good Prompts

• "Extract all product names and prices listed on this page"
• "What are the shipping options and their costs?"
• "List all navigation menu items with their URLs"
• "Is this product currently in stock? Answer yes or no."

Poor Prompts

• "Tell me about this page" (too vague)
• "Everything" (no focus)
• "Data" (meaningless)

Tips

1. Be specific about what data you want
2. Use JSON schema when you need structured output
3. Ask one thing at a time for best accuracy
4. Include format instructions if the default isn't right (e.g., "as a comma-separated list")
5. Use required in schema to ensure fields are always present

Cost Controls

Add cost limits to prevent runaway spending on complex queries.

const result = await client.windows.pageQuery(sessionId, windowId, {
  prompt: 'Extract all pricing tiers',
  configuration: {
    costThresholdCredits: 10,
    timeThresholdSeconds: 30,
  },
});

Best Practices

1. Start with simple queries — add complexity as needed
2. Use JSON schema for any data you'll process programmatically
3. Test prompts on a few pages before scaling with batch operations
4. Set cost/time thresholds for production workloads
5. Combine methods — use scrapeContent for overview, pageQuery for specifics