Give Your Agent Reliable Web Access

The Problem

Agent frameworks make it easy to call a tool and hard to make that tool reliable. A browser-based tool is slow and flaky. A raw HTTP fetch returns a wall of HTML that the model has to summarize before it can act. A homegrown search integration ends up reimplementing half of what a SERP API already does, badly. Every agent team eventually hits the same wall: giving the agent the web is trivial in demos and painful in production because the underlying tool has the wrong shape for an LLM to consume.

The Scavio Solution

Scavio fits the shape of an agent tool exactly. It takes a query, it returns parsed snippets with source URLs, it completes in under two seconds, and the response is small enough to pass back into the context without blowing past the model's limits. You expose a single tool called web_search, the agent calls it whenever it needs the live web, and the tool returns a result that slots into the agent's reasoning chain without further processing. No browser, no HTML, no token-heavy intermediate summarization step.

Who It Is For

Agent builders working with OpenAI Agents SDK, Anthropic tool use, LangGraph, or CrewAI. If your agent needs reliable web access and you are tired of wrapping brittle scrapers, this is the cleanest path.

Key Benefits

• Single tool definition works for OpenAI, Anthropic, and Gemini
• Parsed snippets fit inside sub-thousand-token tool responses
• Predictable sub-two-second latency keeps agent loops snappy
• Supports web, video, and product search from one tool schema
• No browser infrastructure or HTML post-processing required

Python Example

from openai import OpenAI
import requests, json

API_KEY = "your_scavio_api_key"
client = OpenAI()

def web_search(query: str) -> str:
    r = requests.post(
        "https://api.scavio.dev/api/v1/search",
        headers={"x-api-key": API_KEY},
        json={"platform": "google", "query": query},
        timeout=10,
    )
    return json.dumps(r.json().get("organic", [])[:5])

tools = [{
    "type": "function",
    "function": {
        "name": "web_search",
        "description": "Search the live web and return top results",
        "parameters": {"type": "object", "properties": {"query": {"type": "string"}}, "required": ["query"]},
    },
}]

print(web_search("who won euro 2028 qualifiers last night"))

JavaScript Example

const API_KEY = "your_scavio_api_key";

export async function webSearch(query) {
  const r = await fetch("https://api.scavio.dev/api/v1/search", {
    method: "POST",
    headers: {
      "x-api-key": API_KEY,
      "content-type": "application/json",
    },
    body: JSON.stringify({ platform: "google", query }),
  });
  const data = await r.json();
  return JSON.stringify((data.organic ?? []).slice(0, 5));
}

export const webSearchTool = {
  type: "function",
  function: {
    name: "web_search",
    description: "Search the live web and return top results",
    parameters: {
      type: "object",
      properties: { query: { type: "string" } },
      required: ["query"],
    },
  },
};

Platforms Used