HushSearchDocs

Building a Web Agent

Use HushSearch to give your AI agent live web access.

Give your AI agent real-time web access in under 50 lines of Python.

Simple ReAct agent

import anthropic
import requests

client = anthropic.Anthropic()
API_KEY = "hs_live_..."

def search(query: str) -> str:
    r = requests.post(
        "https://api.hushsearch.io/v1/search/",
        headers={"X-API-Key": API_KEY},
        json={"query": query, "num_results": 5}
    )
    results = r.json()["results"]
    return "\n".join(f"- {r['title']}: {r['url']}\n  {r['snippet']}" for r in results)

def extract(url: str) -> str:
    r = requests.post(
        "https://api.hushsearch.io/v1/extract/",
        headers={"X-API-Key": API_KEY},
        json={"url": url}
    )
    return r.json()["markdown_content"][:2000]

tools = [
    {"name": "web_search",    "description": "Search the web",     "input_schema": {"type": "object", "properties": {"query": {"type": "string"}}, "required": ["query"]}},
    {"name": "extract_page",  "description": "Read a web page",    "input_schema": {"type": "object", "properties": {"url":   {"type": "string"}}, "required": ["url"]}},
]

def run_agent(task: str) -> str:
    messages = [{"role": "user", "content": task}]
    while True:
        resp = client.messages.create(model="claude-sonnet-4-6", max_tokens=4096, tools=tools, messages=messages)
        if resp.stop_reason == "end_turn":
            return resp.content[0].text
        tool_results = []
        for block in resp.content:
            if block.type == "tool_use":
                if block.name == "web_search":
                    result = search(block.input["query"])
                elif block.name == "extract_page":
                    result = extract(block.input["url"])
                else:
                    result = "Unknown tool"
                tool_results.append({"type": "tool_result", "tool_use_id": block.id, "content": result})
        messages.append({"role": "assistant", "content": resp.content})
        messages.append({"role": "user", "content": tool_results})

print(run_agent("What are the top 3 LLM providers in 2026 and their pricing?"))

Credits per agent turn

A typical agent run uses:

  • 1-3 search calls = 1-3 credits
  • 2-5 extract calls = 4-10 credits
  • Total: 5-13 credits per research task