🔍Project: Web Research Agent

Build an agent that searches the web, synthesizes results, and produces cited reports

Project: Web Research Agent

In this project, you'll build a smart agent that takes a question, searches the web, gathers information, and writes a comprehensive report with citations using Claude's built-in web search tool.

Main Code

JavaScript
import Anthropic from "@anthropic-ai/sdk";

const client = new Anthropic();

async function researchAgent(question) {
  console.log(`Researching: ${question}\n`);

  const response = await client.messages.create({
    model: "claude-sonnet-4-20250514",
    max_tokens: 4096,
    tools: [{ type: "web_search_20250305" }],
    messages: [{
      role: "user",
      content: `You are a research specialist. Research this topic and write a comprehensive report.

Topic: ${question}

Include: executive summary, key findings, detailed analysis, conclusion, source list with URLs.`,
    }],
  });

  let report = "";
  for (const block of response.content) {
    if (block.type === "text") report += block.text;
  }
  return report;
}

const report = await researchAgent("Latest developments in AI in 2025");
console.log(report);

Python Version

Python
import anthropic

client = anthropic.Anthropic()

def research_agent(question):
    response = client.messages.create(
        model="claude-sonnet-4-20250514",
        max_tokens=4096,
        tools=[{"type": "web_search_20250305"}],
        messages=[{
            "role": "user",
            "content": f"""Research specialist. Research this topic and write a comprehensive report.

Topic: {question}

Include: executive summary, key findings, analysis, conclusion, sources with URLs.""",
        }],
    )

    report = ""
    for block in response.content:
        if block.type == "text":
            report += block.text
    return report

report = research_agent("Latest AI agent frameworks in 2025")
print(report)

Deep Research Agent with Loop

JavaScript
async function deepResearchAgent(question, maxSearches = 5) {
  const messages = [{
    role: "user",
    content: `Research deeply. Search multiple times from different angles.\n\nTopic: ${question}\n\nWrite a detailed report with sources.`,
  }];

  let searches = 0;

  while (searches < maxSearches) {
    const response = await client.messages.create({
      model: "claude-sonnet-4-20250514",
      max_tokens: 4096,
      tools: [{ type: "web_search_20250305" }],
      messages,
    });

    const searchBlocks = response.content.filter(
      (b) => b.type === "server_tool_use" && b.name === "web_search_20250305"
    );
    searches += searchBlocks.length;

    if (response.stop_reason === "end_turn") {
      let report = "";
      for (const block of response.content) {
        if (block.type === "text") report += block.text;
      }
      console.log(`Research complete (${searches} searches)`);
      return report;
    }

    messages.push({ role: "assistant", content: response.content });
    messages.push({ role: "user", content: "Continue researching." });
  }

  return "Max search limit reached";
}

Comparative Research

JavaScript
async function compareTopics(topic1, topic2) {
  return await researchAgent(
    `Compare ${topic1} vs ${topic2}: features, pros, cons, pricing, community. End with a recommendation table.`
  );
}

Saving Reports

JavaScript
import fs from "fs";

async function saveReport(question, outputPath) {
  const report = await researchAgent(question);
  const fullReport = `# Research Report\n## ${question}\n## ${new Date().toISOString().split("T")[0]}\n\n---\n\n${report}`;
  fs.writeFileSync(outputPath, fullReport, "utf-8");
  console.log(`Report saved to: ${outputPath}`);
}

CLI Interface

JavaScript
const question = process.argv.slice(2).join(" ");
if (!question) {
  console.log("Usage: node researcher.js <your research question>");
  process.exit(1);
}

researchAgent(question).then((report) => {
  console.log("\n" + "=".repeat(60) + "\n" + report);
});

Extensions to Try

Comparative research — Compare two topics
Periodic updates — Re-research weekly
PDF export — Convert report to PDF
Web dashboard — Build a research management UI
Source quality scoring — Rate sources by reliability

Next: We'll build a CLI Automation Tool.

Module 8

7/9

⚙️ Project: CLI Automation Tool

Project: AI Data Dashboard 📈

7/9