Add live web search to a Vercel AI SDK agent

Vercel AI SDK's tool() helper and MCP support make it straightforward to give any TypeScript agent live web search — either via an autonomous x402 wallet or a free API key.

Two paths

Path 1: MCP client (recommended for autonomous agents)

npm install ai @ai-sdk/openai zod

import { experimental_createMCPClient, generateText } from 'ai'
import { Experimental_StdioMCPTransport } from 'ai/mcp-stdio'
import { openai } from '@ai-sdk/openai'

async function main() {
  const mcpClient = await experimental_createMCPClient({
    transport: new Experimental_StdioMCPTransport({
      command: 'npx',
      args: ['-y', 'superhighway-mcp'],
      env: {
        AGENT_PRIVATE_KEY: process.env.AGENT_PRIVATE_KEY!,
        X402_NETWORK: 'base',
      },
    }),
  })

  const tools = await mcpClient.tools()

  const { text } = await generateText({
    model: openai('gpt-4o'),
    tools,
    prompt: 'What are the latest developments in AI agent frameworks?',
    maxSteps: 5,
  })

  console.log(text)
  await mcpClient.close()
}

main()

The MCP server exposes all five tools — web_search, news_search, image_search, scrape, research — and each call settles a $0.001–$0.005 USDC payment automatically via x402. See wallet setup or the x402 flow.

Path 2: Typed tool() helpers with an API key

Get a free API key (1,000 calls/month) and wire up typed tools with Zod schemas:

import { generateText, tool } from 'ai'
import { openai } from '@ai-sdk/openai'
import { z } from 'zod'

const BASE = 'https://superhighway.walls.sh'
const KEY = process.env.SUPERHIGHWAY_API_KEY!

const { text } = await generateText({
  model: openai('gpt-4o'),
  tools: {
    webSearch: tool({
      description: 'Search the live web for current information, facts, and events.',
      parameters: z.object({
        query: z.string(),
        limit: z.number().optional(),
      }),
      execute: async ({ query, limit = 5 }) => {
        const res = await fetch(
          `${BASE}/search?q=${encodeURIComponent(query)}&limit=${limit}`,
          { headers: { Authorization: `Bearer ${KEY}` } },
        )
        const data = await res.json()
        return data.results
          .map((r: any) => `${r.title}\n${r.url}\n${r.description}`)
          .join('\n\n')
      },
    }),
    newsSearch: tool({
      description: 'Search recent news articles with published dates.',
      parameters: z.object({ query: z.string() }),
      execute: async ({ query }) => {
        const res = await fetch(
          `${BASE}/news?q=${encodeURIComponent(query)}`,
          { headers: { Authorization: `Bearer ${KEY}` } },
        )
        const data = await res.json()
        return data.results
          .map((r: any) => `${r.title} (${r.publishedDate ?? ''})\n${r.url}\n${r.description}`)
          .join('\n\n')
      },
    }),
    research: tool({
      description: 'Search + read the top matching pages in one call. Use for deep synthesis.',
      parameters: z.object({
        query: z.string(),
        pages: z.number().optional(),
      }),
      execute: async ({ query, pages = 2 }) => {
        const res = await fetch(
          `${BASE}/research?q=${encodeURIComponent(query)}&pages=${pages}`,
          { headers: { Authorization: `Bearer ${KEY}` } },
        )
        const data = await res.json()
        return data.pages
          .filter((p: any) => !p.error)
          .map((p: any) => `## ${p.title}\n${p.url}\n\n${p.markdown.slice(0, 3000)}`)
          .join('\n\n---\n\n')
      },
    }),
  },
  prompt: 'Summarize the top AI news from the last 48 hours.',
  maxSteps: 5,
})

console.log(text)

Each tool() is typed end-to-end with Zod: the SDK validates inputs before calling execute and passes the return value back to the model as a tool result. Add only what the agent needs — webSearch + newsSearch for real-time Q&A, research for multi-page synthesis.

Which tools are available

Full API reference: /openapi.json. Get a free API key or learn the x402 wallet flow.