Add ___domain filtering options to web_search tool in Responses API

[jay-dhamale]

Feature Request

Add the ability to filter web search results by including or excluding specific domains/websites in the OpenAI Responses API web_search tool.

Problem Statement

Currently, the OpenAI Responses API web_search tool configuration only supports basic parameters:

response = client.responses.create(
    model="gpt-4o",
    tools=[
        {
            "type": "web_search_preview",
            "user_location": {
                "type": "approximate",
                "country": "US",
                "city": "San Francisco",
            },
        }
    ],
    input="Search query here",
)

This limits users who need to:

• Focus searches on specific trusted sources
• Exclude known unreliable or irrelevant domains
• Customize search results for specific use cases (e.g., academic research, official sources only)

Proposed Solution

Add two new optional parameters to the web_search tool configuration, similar to Perplexity AI's implementation:

1. include_domains: List of domains to limit search results to
2. exclude_domains: List of domains to exclude from search results

Example Implementation

Current OpenAI Implementation:

response = client.responses.create(
    model="gpt-4o",
    tools=[
        {
            "type": "web_search_preview",
            "user_location": {
                "type": "approximate",
                "country": "US",
            },
        }
    ],
    input="Latest AI research papers",
)

Proposed Enhancement:

response = client.responses.create(
    model="gpt-4o",
    tools=[
        {
            "type": "web_search_preview",
            "user_location": {
                "type": "approximate",
                "country": "US",
            },
            # New ___domain filtering parameters
            "include_domains": ["arxiv.org", "openai.com", "nature.com"],
            "exclude_domains": ["medium.com", "reddit.com"]
        }
    ],
    input="Latest AI research papers",
)

Reference: Perplexity AI Implementation

Perplexity's API already supports this functionality:

import requests

response = requests.post(
    "https://api.perplexity.ai/search",
    headers={
        "Authorization": "Bearer YOUR_API_KEY",
        "Content-Type": "application/json"
    },
    json={
        "query": "machine learning research",
        "include_domains": ["arxiv.org", "ieee.org", "acm.org"],
        "exclude_domains": ["blogspot.com", "wordpress.com"]
    }
)

Use Cases

1. Academic Research: Include only .edu domains and academic publishers
2. Official Information: Focus on government (.gov) and organizational (.org) domains
3. Technical Documentation: Include official documentation sites, exclude forums
4. News Aggregation: Include trusted news sources, exclude tabloids
5. Product Research: Include official vendor sites, exclude affiliate spam

Implementation Considerations

• Both parameters should be optional to maintain backward compatibility
• Support for wildcard patterns (e.g., *.edu, *.gov)
• Consider adding validation for ___domain format
• Document any limitations on the number of domains that can be filtered
• Ensure the filtering happens at the search API level for efficiency

Benefits

• More precise and relevant search results
• Better control over information sources
• Reduced noise from unreliable sources
• Improved efficiency by filtering early in the search process
• Feature parity with competing APIs like Perplexity

Additional Considerations

The web_search tool could also benefit from:

• A sites parameter to search within specific sites only (similar to Google's site: operator)
• Support for excluding specific URL patterns, not just domains
• Option to prioritize certain domains in results ranking

This enhancement would significantly improve the utility of the web_search tool for developers building applications that require high-quality, ___domain-specific information retrieval.