How to integrate Browserbase tool MCP with Vercel AI SDK v6

Connect Vercel AI SDK v6 to Browserbase tool MCP. Start a new headless browser session now, download all artifacts from last session, and more using natural language, with authentication handled for you.

Browserbase tool logoBrowserbase tool
Api Key

Browserbase is a serverless platform for running and managing headless browsers at scale. It lets you automate and monitor browser tasks with Playwright, Puppeteer, and Selenium—no infrastructure headaches.

64 Tools

Introduction

This guide walks you through connecting Browserbase tool to Vercel AI SDK v6 using the Composio tool router. By the end, you'll have a working Browserbase tool agent that can create and manage browser sessions, run Browserbase agents, invoke functions, fetch pages, and download session replays and logs through natural language commands.

This guide will help you understand how to give your Vercel AI SDK agent real control over a Browserbase tool account through Composio's Browserbase tool MCP server.

Before we dive in, let's take a quick look at the key ideas and tools involved.

For exact tool slugs and parameter tables, see the tool catalog in the Composio docs.

Also integrate Browserbase tool with

TL;DR

Here's what you'll learn:
  • How to set up and configure a Vercel AI SDK agent with Browserbase tool integration
  • Using Composio's Tool Router to dynamically load and access Browserbase tool tools
  • Creating an MCP client connection using HTTP transport
  • Building an interactive CLI chat interface with conversation history management
  • Handling tool calls and results within the Vercel AI SDK framework

What is Vercel AI SDK?

The Vercel AI SDK is a TypeScript library for building AI-powered applications. It provides tools for creating agents that can use external services and maintain conversation state.

Key features include:

  • streamText: Core function for streaming responses with real-time tool support
  • MCP Client: Built-in support for Model Context Protocol via @ai-sdk/mcp
  • Step Counting: Control multi-step tool execution with stopWhen: stepCountIs()
  • OpenAI Provider: Native integration with OpenAI models

What is the Browserbase tool MCP server, and what's possible with it?

The Browserbase MCP server is an implementation of the Model Context Protocol that connects your AI agent and assistants like Claude, Cursor, etc directly to your Browserbase account. It provides structured and secure access to your browser sessions, contexts, projects, recordings, logs, files, and extensions, so your agent can create sessions, manage reusable contexts, inspect browser activity, retrieve recordings, and monitor project usage on your behalf.

  • Browser session management: Have your agent create isolated browser sessions, check their status, list recent sessions, and release running sessions when work is complete.
  • Reusable browser contexts: Let the agent create, inspect, and update browser contexts that preserve settings across multiple sessions.
  • Session monitoring and debugging: Direct your agent to retrieve live debugging access, review captured network activity, and inspect session logs.
  • Recordings and session artifacts: Instruct your agent to retrieve session replays, request downloadable recordings, and download generated session files.
  • Projects, files, and extensions: Have your agent review projects and usage, upload files for browser forms, and upload or inspect browser extensions.

What is the Composio tool router, and how does it fit here?

What is Composio SDK?

Composio's Composio SDK helps agents find the right tools for a task at runtime. You can plug in multiple toolkits (like Gmail, HubSpot, and GitHub), and the agent will identify the relevant app and action to complete multi-step workflows. This can reduce token usage and improve the reliability of tool calls. Read more here: Getting started with Composio SDK

The tool router generates a secure MCP URL that your agents can access to perform actions.

How the Composio SDK works

The Composio SDK follows a three-phase workflow:

  1. Discovery: Searches for tools matching your task and returns relevant toolkits with their details.
  2. Authentication: Checks for active connections. If missing, creates an auth config and returns a connection URL via Auth Link.
  3. Execution: Executes the action using the authenticated connection.

Step-by-step Guide

Step by step09 STEPS
1

Prerequisites

Before you begin, make sure you have:
  • Node.js and npm installed
  • A Composio account with API key
  • An OpenAI API key
2

Getting API Keys for OpenAI and Composio

OpenAI API Key
  • Go to the OpenAI dashboard and create an API key. You'll need credits to use the models, or you can connect to another model provider.
  • Keep the API key safe.
Composio API Key
  • Log in to the Composio dashboard.
  • Navigate to your API settings and generate a new API key.
  • Store this key securely as you'll need it for authentication.
3

Install required dependencies

bash
npm install @ai-sdk/openai @ai-sdk/mcp @composio/core ai dotenv

First, install the necessary packages for your project.

What you're installing:

  • @ai-sdk/openai: Vercel AI SDK's OpenAI provider
  • @ai-sdk/mcp: MCP client for Vercel AI SDK
  • @composio/core: Composio SDK for tool integration
  • ai: Core Vercel AI SDK
  • dotenv: Environment variable management
4

Set up environment variables

bash
OPENAI_API_KEY=your_openai_api_key_here
COMPOSIO_API_KEY=your_composio_api_key_here
COMPOSIO_USER_ID=your_user_id_here

Create a .env file in your project root.

What's needed:

  • OPENAI_API_KEY: Your OpenAI API key for GPT model access
  • COMPOSIO_API_KEY: Your Composio API key for tool access
  • COMPOSIO_USER_ID: A unique identifier for the user session
5

Import required modules and validate environment

typescript
import "dotenv/config";
import { openai } from "@ai-sdk/openai";
import { Composio } from "@composio/core";
import * as readline from "readline";
import { streamText, type ModelMessage, stepCountIs } from "ai";
import { createMCPClient } from "@ai-sdk/mcp";

const composioAPIKey = process.env.COMPOSIO_API_KEY;
const composioUserID = process.env.COMPOSIO_USER_ID;

if (!process.env.OPENAI_API_KEY) throw new Error("OPENAI_API_KEY is not set");
if (!composioAPIKey) throw new Error("COMPOSIO_API_KEY is not set");
if (!composioUserID) throw new Error("COMPOSIO_USER_ID is not set");

const composio = new Composio({
  apiKey: composioAPIKey,
});
What's happening:
  • We're importing all necessary libraries including Vercel AI SDK's OpenAI provider and Composio
  • The dotenv/config import automatically loads environment variables
  • The MCP client import enables connection to Composio's tool server
6

Create Tool Router session and initialize MCP client

typescript
async function main() {
  // Create a tool router session for the user
  const session = await composio.create(composioUserID!, {
    toolkits: ["browserbase_tool"],
  });

  const mcpUrl = session.mcp.url;
What's happening:
  • We're creating a Tool Router session that gives your agent access to Browserbase tool tools
  • The create method takes the user ID and specifies which toolkits should be available
  • The returned mcp object contains the URL and authentication headers needed to connect to the MCP server
  • This session provides access to all Browserbase tool-related tools through the MCP protocol
7

Connect to MCP server and retrieve tools

typescript
const mcpClient = await createMCPClient({
  transport: {
    type: "http",
    url: mcpUrl,
    headers: session.mcp.headers, // Authentication headers for the Composio MCP server
  },
});

const tools = await mcpClient.tools();
What's happening:
  • We're creating an MCP client that connects to our Composio Tool Router session via HTTP
  • The mcp.url provides the endpoint, and mcp.headers contains authentication credentials
  • The type: "http" is important - Composio requires HTTP transport
  • tools() retrieves all available Browserbase tool tools that the agent can use
8

Initialize conversation and CLI interface

typescript
let messages: ModelMessage[] = [];

console.log("Chat started! Type 'exit' or 'quit' to end the conversation.\n");
console.log(
  "Ask any questions related to browserbase_tool, like summarize my last 5 emails, send an email, etc... :)))\n",
);

const rl = readline.createInterface({
  input: process.stdin,
  output: process.stdout,
  prompt: "> ",
});

rl.prompt();
What's happening:
  • We initialize an empty messages array to maintain conversation history
  • A readline interface is created to accept user input from the command line
  • Instructions are displayed to guide the user on how to interact with the agent
9

Handle user input and stream responses with real-time tool feedback

typescript
rl.on("line", async (userInput: string) => {
  const trimmedInput = userInput.trim();

  if (["exit", "quit", "bye"].includes(trimmedInput.toLowerCase())) {
    console.log("\nGoodbye!");
    rl.close();
    process.exit(0);
  }

  if (!trimmedInput) {
    rl.prompt();
    return;
  }

  messages.push({ role: "user", content: trimmedInput });
  console.log("\nAgent is thinking...\n");

  try {
    const stream = streamText({
      model: openai("gpt-5"),
      messages,
      tools,
      toolChoice: "auto",
      stopWhen: stepCountIs(10),
      onStepFinish: (step) => {
        for (const toolCall of step.toolCalls) {
          console.log(`[Using tool: ${toolCall.toolName}]`);
          }
          if (step.toolCalls.length > 0) {
            console.log(""); // Add space after tool calls
          }
        },
      });

      for await (const chunk of stream.textStream) {
        process.stdout.write(chunk);
      }

      console.log("\n\n---\n");

      // Get final result for message history
      const response = await stream.response;
      if (response?.messages?.length) {
        messages.push(...response.messages);
      }
    } catch (error) {
      console.error("\nAn error occurred while talking to the agent:");
      console.error(error);
      console.log(
        "\nYou can try again or restart the app if it keeps happening.\n",
      );
    } finally {
      rl.prompt();
    }
  });

  rl.on("close", async () => {
    await mcpClient.close();
    console.log("\n👋 Session ended.");
    process.exit(0);
  });
}

main().catch((err) => {
  console.error("Fatal error:", err);
  process.exit(1);
});
What's happening:
  • We use streamText instead of generateText to stream responses in real-time
  • toolChoice: "auto" allows the model to decide when to use Browserbase tool tools
  • stopWhen: stepCountIs(10) allows up to 10 steps for complex multi-tool operations
  • onStepFinish callback displays which tools are being used in real-time
  • We iterate through the text stream to create a typewriter effect as the agent responds
  • The complete response is added to conversation history to maintain context
  • Errors are caught and displayed with helpful retry suggestions

Complete Code

Here's the complete code to get you started with Browserbase tool and Vercel AI SDK:

typescript
import "dotenv/config";
import { openai } from "@ai-sdk/openai";
import { Composio } from "@composio/core";
import * as readline from "readline";
import { streamText, type ModelMessage, stepCountIs } from "ai";
import { createMCPClient } from "@ai-sdk/mcp";

const composioAPIKey = process.env.COMPOSIO_API_KEY;
const composioUserID = process.env.COMPOSIO_USER_ID;

if (!process.env.OPENAI_API_KEY) throw new Error("OPENAI_API_KEY is not set");
if (!composioAPIKey) throw new Error("COMPOSIO_API_KEY is not set");
if (!composioUserID) throw new Error("COMPOSIO_USER_ID is not set");

const composio = new Composio({
  apiKey: composioAPIKey,
});

async function main() {
  // Create a tool router session for the user
  const session = await composio.create(composioUserID!, {
    toolkits: ["browserbase_tool"],
  });

  const mcpUrl = session.mcp.url;

  const mcpClient = await createMCPClient({
    transport: {
      type: "http",
      url: mcpUrl,
      headers: session.mcp.headers, // Authentication headers for the Composio MCP server
    },
  });

  const tools = await mcpClient.tools();

  let messages: ModelMessage[] = [];

  console.log("Chat started! Type 'exit' or 'quit' to end the conversation.\n");
  console.log(
    "Ask any questions related to browserbase_tool, like summarize my last 5 emails, send an email, etc... :)))\n",
  );

  const rl = readline.createInterface({
    input: process.stdin,
    output: process.stdout,
    prompt: "> ",
  });

  rl.prompt();

  rl.on("line", async (userInput: string) => {
    const trimmedInput = userInput.trim();

    if (["exit", "quit", "bye"].includes(trimmedInput.toLowerCase())) {
      console.log("\nGoodbye!");
      rl.close();
      process.exit(0);
    }

    if (!trimmedInput) {
      rl.prompt();
      return;
    }

    messages.push({ role: "user", content: trimmedInput });
    console.log("\nAgent is thinking...\n");

    try {
      const stream = streamText({
        model: openai("gpt-5"),
        messages,
        tools,
        toolChoice: "auto",
        stopWhen: stepCountIs(10),
        onStepFinish: (step) => {
          for (const toolCall of step.toolCalls) {
            console.log(`[Using tool: ${toolCall.toolName}]`);
          }
          if (step.toolCalls.length > 0) {
            console.log(""); // Add space after tool calls
          }
        },
      });

      for await (const chunk of stream.textStream) {
        process.stdout.write(chunk);
      }

      console.log("\n\n---\n");

      // Get final result for message history
      const response = await stream.response;
      if (response?.messages?.length) {
        messages.push(...response.messages);
      }
    } catch (error) {
      console.error("\nAn error occurred while talking to the agent:");
      console.error(error);
      console.log(
        "\nYou can try again or restart the app if it keeps happening.\n",
      );
    } finally {
      rl.prompt();
    }
  });

  rl.on("close", async () => {
    await mcpClient.close();
    console.log("\n👋 Session ended.");
    process.exit(0);
  });
}

main().catch((err) => {
  console.error("Fatal error:", err);
  process.exit(1);
});

Conclusion

You've successfully built a Browserbase tool agent using the Vercel AI SDK with streaming capabilities! This implementation provides a powerful foundation for building AI applications with natural language interfaces and real-time feedback.

Key features of this implementation:

  • Real-time streaming responses for a better user experience with typewriter effect
  • Live tool execution feedback showing which tools are being used as the agent works
  • Dynamic tool loading through Composio's Tool Router with secure authentication
  • Multi-step tool execution with configurable step limits (up to 10 steps)
  • Comprehensive error handling for robust agent execution
  • Conversation history maintenance for context-aware responses

You can extend this further by adding custom error handling, implementing specific business logic, or integrating additional Composio toolkits to create multi-app workflows.
TOOLS

Supported Tools

Every Browserbase tool action and event your agent gets out of the box.

Attach Browserbase Function Secret

Attach an existing project secret to a Function by UUID.

Update Browser Context

Tool to update a specific browser context.

Create Browserbase Agent

Create a reusable browser agent with a name and optional system prompt and result schema.

Create Browser Context

Create a reusable browser context; the API key can select the project.

Create Browser Session

Create an isolated Browserbase session and return its connection details.

Create Browserbase Webhook

Register a publicly reachable HTTPS endpoint for project event deliveries.

Delete Browserbase Agent

Delete a reusable browser agent by its agentId.

Delete a browser context

Tool to delete a browser context and all its stored data (cookies, localStorage, etc.).

Delete Download

Delete a download file from storage and mark it as deleted.

Delete a browser extension

Tool to delete an uploaded browser extension by its ID.

Delete Session Downloads

Tool to delete all file downloads from a specific browser session.

Delete Browserbase Webhook

Delete a webhook, immediately stopping deliveries and retiring its secret.

Detach Browserbase Function Secret

Detach a project secret from a Function by UUID.

Download Session Artifacts

Download a completed session's generated files as a ZIP archive.

Fetch Browserbase Page

Fetch a page and return its content, headers, and metadata.

Get Browserbase Agent

Retrieve a reusable browser agent by its agentId.

Get Certificate

Retrieve a Browserbase certificate by its ID.

Get Browser Context

Return metadata for a browser context by ID.

Get Download

Get a single download by ID as metadata or a downloadable file object.

Retrieve a browser extension

Tool to retrieve details of a specific browser extension.

Get Browserbase Function

Retrieve a Browserbase function by its ID.

Get Browserbase Function Build

Retrieve a Browserbase function build by its build ID.

Get Browserbase Function Build Logs

Retrieve Browserbase function build logs by the build ID.

Get Browserbase Function Version

Retrieve a Browserbase function version by its version ID.

Get Browserbase Invocation

Retrieve a Browserbase function invocation by its invocation ID.

Get Browserbase Invocation Logs

Retrieve Browserbase function invocation logs by the invocation ID.

Get Browserbase Keypair

Retrieve the project's public HPKE encryption key and keypair ID.

Get Browserbase Project

Return project metadata, default timeout, and assigned concurrency.

Get project usage statistics

Tool to retrieve usage statistics for a project including browser minutes and proxy bytes consumed.

Get Browserbase Agent Run

Retrieve a browser agent run's current status, result, and session details.

Get Browserbase Secret

Retrieve a project secret's ID and name by UUID, without its secret value.

Get Browser Session

Return status, connection details, and timestamps for a browser session.

Get Session Debug URLs

Return live debugger and WebSocket URLs for a running browser session.

Get Session Logs

Return network request and response events captured during a browser session.

Get Session Replay

Return HLS replay metadata and page playlist paths for a recorded session.

Get Session Replay Page

Return the HLS M3U8 playlist for one recorded replay page.

Get Browserbase Webhook

Retrieve a single webhook by ID, without its signing secret.

Invoke Browserbase Function

Invoke a deployed Browserbase function asynchronously.

List Browserbase Agents

List one page of reusable browser agents across your account.

List Certificates

List Browserbase certificates with their IDs, timestamps, and project IDs.

List Browserbase Downloads

List one page of download metadata for a required browser session.

List Browserbase Function Builds

List one page of Browserbase function builds, optionally filtered by status.

List Browserbase Functions

List one page of Browserbase functions and the total number available.

List Browserbase Function Secrets

List one page of secret IDs and names attached to a Browserbase function, optionally filtered by creation time.

List Browserbase Function Versions

List one page of versions for a Browserbase function by its UUID.

List Browserbase Projects

List projects with their default timeout and assigned concurrency.

List Recording Downloads

Poll per-page MP4 assembly after requesting recording downloads until every page is COMPLETED or FAILED.

List Browserbase Run Messages

List a page of a browser agent run's messages, oldest first, with assistant or tool roles and text or typed content parts.

List Browserbase Agent Runs

List a page of browser agent runs across your account.

List Browserbase Secrets

List one page of project secret IDs and names, optionally filtered by creation time.

List Browser Sessions

List browser sessions, optionally filtered by status or user metadata.

List Browserbase Function Version Invocations

List one page of invocations for a Browserbase function version by its UUID.

List Browserbase Webhooks

List one page of the project's webhooks, newest first, without signing secrets.

Release Browser Session

Irreversibly request release of a running browser session, ending it before its configured timeout to avoid further charges.

Request Recording Downloads

Start asynchronous MP4 assembly for every recorded page of a completed session, then poll List Recording Downloads until every page is COMPLETED or FAILED.

Resume Browserbase Agent Run

Resume only a PAUSED browser agent run with optional additional input.

Run Browserbase Agent

Start a browser agent run to complete a task using web search and browser tools.

Stop Browserbase Agent Run

Request that an in-progress browser agent run stop.

Update Browserbase Agent

Update a reusable browser agent's name, system prompt, or result schema.

Update Browserbase Webhook

Update a webhook's delivery URL, subscribed event types, or both.

Upload Browser Extension

Tool to upload a browser extension for use in sessions.

Upload Session File

Upload one file so browser interactions can select it in a file input field.

Upload Session Files

Upload multiple files to one browser session for file-input interactions.

Web Search

Perform a web search and return structured results.

FAQ

Frequently asked questions

Start with Browserbase tool.It takes 30 seconds.

Managed auth, hosted MCP servers, and every Browserbase tool tool your agent needs.Free to start.

Start building