How to integrate Google cloud vision MCP with LlamaIndex

This guide walks you through connecting Google cloud vision to LlamaIndex using the Composio tool router. By the end, you'll have a working Google cloud vision agent that can bulk import product images from gcs csv, list all vision ai service locations, create a new product for image recognition through natural language commands. This guide will help you understand how to give your LlamaIndex agent real control over a Google cloud vision account through Composio's Google cloud vision MCP server. Before we dive in, let's take a quick look at the key ideas and tools involved.

Google cloud vision logoGoogle cloud vision
Api Key

Google Cloud Vision API adds advanced image analysis—like labeling, OCR, and detection—to apps. It helps you extract structured data and insights from images at scale.

29 Tools

Introduction

This guide walks you through connecting Google cloud vision to LlamaIndex using the Composio tool router. By the end, you'll have a working Google cloud vision agent that can bulk import product images from gcs csv, list all vision ai service locations, create a new product for image recognition through natural language commands.

This guide will help you understand how to give your LlamaIndex agent real control over a Google cloud vision account through Composio's Google cloud vision MCP server.

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

Also integrate Google cloud vision with

TL;DR

Here's what you'll learn:
  • Set your OpenAI and Composio API keys
  • Install LlamaIndex and Composio packages
  • Create a Composio Tool Router session for Google cloud vision
  • Connect LlamaIndex to the Google cloud vision MCP server
  • Build a Google cloud vision-powered agent using LlamaIndex
  • Interact with Google cloud vision through natural language

What is LlamaIndex?

LlamaIndex is a data framework for building LLM applications. It provides tools for connecting LLMs to external data sources and services through agents and tools.

Key features include:

  • ReAct Agent: Reasoning and acting pattern for tool-using agents
  • MCP Tools: Native support for Model Context Protocol
  • Context Management: Maintain conversation context across interactions
  • Async Support: Built for async/await patterns

What is the Google cloud vision MCP server, and what's possible with it?

The Google cloud vision MCP server is an implementation of the Model Context Protocol that connects your AI agent and assistants like Claude, Cursor, etc directly to your Google Cloud Vision account. It provides structured and secure access to your image analysis resources, so your agent can perform actions like registering products, managing reference images, listing endpoints, and automating large-scale image operations on your behalf.

  • Product and reference image management: Easily create new products and add reference images for visual search, enabling your agent to organize and expand your vision datasets effortlessly.
  • Bulk import and product set operations: Let your agent import large numbers of reference images into product sets from Cloud Storage CSV files, streamlining dataset curation at scale.
  • Automated product cleanup and deletion: Direct your agent to purge unused or orphan products from your project, keeping your cloud resources tidy without manual effort.
  • Location and endpoint discovery: Quickly list available Vision AI service locations and existing IndexEndpoints, making it easy for your agent to select optimal regions and manage deployment targets.
  • Vision API operation tracking: Retrieve and review ongoing or past Vision API operations, so your agent can monitor processing jobs and ensure workflow transparency.

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 step10 STEPS
1

Prerequisites

Before you begin, make sure you have:
  • Python 3.8/Node 16 or higher installed
  • A Composio account with the API key
  • An OpenAI API key
  • A Google cloud vision account and project
  • Basic familiarity with async Python/Typescript
2

Getting API Keys for OpenAI, Composio, and Google cloud vision

OpenAI API key (OPENAI_API_KEY)
  • Go to the OpenAI dashboard
  • Create an API key if you don't have one
  • Assign it to OPENAI_API_KEY in .env
Composio API key and user ID
  • Log into the Composio dashboard
  • Copy your API key from Settings
    • Use this as COMPOSIO_API_KEY
  • Pick a stable user identifier (email or ID)
    • Use this as COMPOSIO_USER_ID
3

Installing dependencies

npm install @composio/llamaindex @llamaindex/openai @llamaindex/tools @llamaindex/workflow dotenv

Create a new Typescript project and install the necessary dependencies:

  • @composio/llamaindex: Composio's LlamaIndex integration
  • @llamaindex/openai: OpenAI LLM integration
  • @llamaindex/tools: MCP client for LlamaIndex
  • @llamaindex/workflow: Workflow framework for LlamaIndex
  • dotenv: Environment variable management
4

Set environment variables

bash
OPENAI_API_KEY=your-openai-api-key
COMPOSIO_API_KEY=your-composio-api-key
COMPOSIO_USER_ID=your-user-id

Create a .env file in your project root:

These credentials will be used to:

  • Authenticate with OpenAI's GPT-5 model
  • Connect to Composio's Tool Router
  • Identify your Composio user session for Google cloud vision access
5

Import modules

import "dotenv/config";
import readline from "node:readline/promises";
import { stdin as input, stdout as output } from "node:process";

import { Composio } from "@composio/core";

import { mcp } from "@llamaindex/tools";
import { agent as createAgent } from "@llamaindex/workflow";
import { openai } from "@llamaindex/openai";

dotenv.config();

Create a new file called google cloud vision_llamaindex_agent.ts and import the required modules:

Key imports:

  • dotenv.config loads .env at runtime
  • readline gives us a simple CLI chat loop
  • Composio is the main Composio SDK client
  • mcp connects to an MCP endpoint
  • createAgent builds a LlamaIndex agent
  • openai configures the LLM backend
6

Load environment variables and initialize Composio

const OPENAI_API_KEY = process.env.OPENAI_API_KEY;
const COMPOSIO_API_KEY = process.env.COMPOSIO_API_KEY;
const COMPOSIO_USER_ID = process.env.COMPOSIO_USER_ID;

if (!OPENAI_API_KEY) throw new Error("OPENAI_API_KEY is not set");
if (!COMPOSIO_API_KEY) throw new Error("COMPOSIO_API_KEY is not set");
if (!COMPOSIO_USER_ID) throw new Error("COMPOSIO_USER_ID is not set");

What's happening:

This ensures missing credentials cause early, clear errors before the agent attempts to initialise.

7

Create a Tool Router session and build the agent function

async function buildAgent() {

  console.log(`Initializing Composio client...${COMPOSIO_USER_ID!}...`);
  console.log(`COMPOSIO_USER_ID: ${COMPOSIO_USER_ID!}...`);

  const composio = new Composio({
    apiKey: COMPOSIO_API_KEY,
    provider: new LlamaindexProvider(),
  });

  const session = await composio.create(
    COMPOSIO_USER_ID!,
    {
      toolkits: ["google_cloud_vision"],
    },
  );

  const mcpUrl = session.mcp.url;
  console.log(`Composio Tool Router MCP URL: ${mcpUrl}`);

  const server = mcp({
    url: mcpUrl,
    clientName: "composio_tool_router_with_llamaindex",
    requestInit: {
      headers: {
        "x-api-key": COMPOSIO_API_KEY!,
      },
    },
    // verbose: true,
  });

  const tools = await server.tools();

  const llm = openai({ apiKey: OPENAI_API_KEY, model: "gpt-5" });

  const agent = createAgent({
    name: "composio_tool_router_with_llamaindex",
        description : "An agent that uses Composio Tool Router MCP tools to perform actions.",
    systemPrompt:
      "You are a helpful assistant connected to Composio Tool Router."+
"Use the available tools to answer user queries and perform Google cloud vision actions." ,
    llm,
    tools,
  });

  return agent;
}

What's happening here:

  • We create a Composio client using your API key and configure it with the LlamaIndex provider
  • We then create a tool router MCP session for your user, specifying the toolkits we want to use (in this case, google cloud vision)
  • The session returns an MCP HTTP endpoint URL that acts as a gateway to all your configured tools
  • LlamaIndex will connect to this endpoint to dynamically discover and use the available Google cloud vision tools.
  • The MCP tools are mapped to LlamaIndex-compatible tools and plug them into the Agent.
8

Create an interactive chat loop

async function chatLoop(agent: ReturnType<typeof createAgent>) {
  const rl = readline.createInterface({ input, output });

  console.log("Type 'quit' or 'exit' to stop.");

  while (true) {
    let userInput: string;

    try {
      userInput = (await rl.question("\nYou: ")).trim();
    } catch {
      console.log("\nAgent: Bye!");
      break;
    }

    if (!userInput) {
      continue;
    }

    const lower = userInput.toLowerCase();
    if (lower === "quit" || lower === "exit") {
      console.log("Agent: Bye!");
      break;
    }

    try {
      process.stdout.write("Agent: ");

      const stream = agent.runStream(userInput);
      let finalResult: any = null;

      for await (const event of stream) {
        // The event.data contains the streamed content
        const data: any = event.data;

        // Check for streaming delta content
        if (data?.delta) {
          process.stdout.write(data.delta);
        }

        // Store final result for fallback
        if (data?.result || data?.message) {
          finalResult = data;
        }
      }

      // If no streaming happened, show the final result
      if (finalResult) {
        const answer =
          finalResult.result ??
          finalResult.message?.content ??
          finalResult.message ??
          "";
        if (answer && typeof answer === "string" && !answer.includes("[object")) {
          process.stdout.write(answer);
        }
      }

      console.log(); // New line after streaming completes
    } catch (err: any) {
      console.error("\nAgent error:", err?.message ?? err);
    }
  }

  rl.close();
}

What's happening:

  • We're creating a direct terminal interface to chat with Google cloud vision
  • The LLM's responses are streamed to the CLI for faster interaction.
  • The agent uses context to maintain conversation history
  • The agent processes the request, selects appropriate Google cloud vision tools, and returns a result
  • We extract the answer from the result data structure and display it to the user
  • You can type 'quit' or 'exit' to stop the chat loop gracefully
  • Agent responses and any errors are streamed in a clear, readable format
9

Define the main entry point

async function main() {
  try {
    const agent = await buildAgent();
    await chatLoop(agent);
  } catch (err) {
    console.error("Failed to start agent:", err);
    process.exit(1);
  }
}

main();

What's happening here:

  • We're orchestrating the entire application flow
  • The agent gets built with proper error handling
  • Then we kick off the interactive chat loop so you can start talking to Google cloud vision
10

Run the agent

npx ts-node llamaindex-agent.ts

When prompted, authenticate and authorise your agent with Google cloud vision, then start asking questions.

Complete Code

Here's the complete code to get you started with Google cloud vision and LlamaIndex:

import "dotenv/config";
import readline from "node:readline/promises";
import { stdin as input, stdout as output } from "node:process";

import { Composio } from "@composio/core";
import { LlamaindexProvider } from "@composio/llamaindex";

import { mcp } from "@llamaindex/tools";
import { agent as createAgent } from "@llamaindex/workflow";
import { openai } from "@llamaindex/openai";

dotenv.config();

const OPENAI_API_KEY = process.env.OPENAI_API_KEY;
const COMPOSIO_API_KEY = process.env.COMPOSIO_API_KEY;
const COMPOSIO_USER_ID = process.env.COMPOSIO_USER_ID;

if (!OPENAI_API_KEY) {
    throw new Error("OPENAI_API_KEY is not set in the environment");
  }
if (!COMPOSIO_API_KEY) {
    throw new Error("COMPOSIO_API_KEY is not set in the environment");
  }
if (!COMPOSIO_USER_ID) {
    throw new Error("COMPOSIO_USER_ID is not set in the environment");
  }

async function buildAgent() {

  console.log(`Initializing Composio client...${COMPOSIO_USER_ID!}...`);
  console.log(`COMPOSIO_USER_ID: ${COMPOSIO_USER_ID!}...`);

  const composio = new Composio({
    apiKey: COMPOSIO_API_KEY,
    provider: new LlamaindexProvider(),
  });

  const session = await composio.create(
    COMPOSIO_USER_ID!,
    {
      toolkits: ["google_cloud_vision"],
    },
  );

  const mcpUrl = session.mcp.url;
  console.log(`Composio Tool Router MCP URL: ${mcpUrl}`);

  const server = mcp({
    url: mcpUrl,
    clientName: "composio_tool_router_with_llamaindex",
    requestInit: {
      headers: {
        "x-api-key": COMPOSIO_API_KEY!,
      },
    },
    // verbose: true,
  });

  const tools = await server.tools();

  const llm = openai({ apiKey: OPENAI_API_KEY, model: "gpt-5" });

  const agent = createAgent({
    name: "composio_tool_router_with_llamaindex",
    description:
      "An agent that uses Composio Tool Router MCP tools to perform actions.",
    systemPrompt:
      "You are a helpful assistant connected to Composio Tool Router."+
"Use the available tools to answer user queries and perform Google cloud vision actions." ,
    llm,
    tools,
  });

  return agent;
}

async function chatLoop(agent: ReturnType<typeof createAgent>) {
  const rl = readline.createInterface({ input, output });

  console.log("Type 'quit' or 'exit' to stop.");

  while (true) {
    let userInput: string;

    try {
      userInput = (await rl.question("\nYou: ")).trim();
    } catch {
      console.log("\nAgent: Bye!");
      break;
    }

    if (!userInput) {
      continue;
    }

    const lower = userInput.toLowerCase();
    if (lower === "quit" || lower === "exit") {
      console.log("Agent: Bye!");
      break;
    }

    try {
      process.stdout.write("Agent: ");

      const stream = agent.runStream(userInput);
      let finalResult: any = null;

      for await (const event of stream) {
        // The event.data contains the streamed content
        const data: any = event.data;

        // Check for streaming delta content
        if (data?.delta) {
          process.stdout.write(data.delta);
        }

        // Store final result for fallback
        if (data?.result || data?.message) {
          finalResult = data;
        }
      }

      // If no streaming happened, show the final result
      if (finalResult) {
        const answer =
          finalResult.result ??
          finalResult.message?.content ??
          finalResult.message ??
          "";
        if (answer && typeof answer === "string" && !answer.includes("[object")) {
          process.stdout.write(answer);
        }
      }

      console.log(); // New line after streaming completes
    } catch (err: any) {
      console.error("\nAgent error:", err?.message ?? err);
    }
  }

  rl.close();
}

async function main() {
  try {
    const agent = await buildAgent();
    await chatLoop(agent);
  } catch (err: any) {
    console.error("Failed to start agent:", err?.message ?? err);
    process.exit(1);
  }
}

main();

Conclusion

You've successfully connected Google cloud vision to LlamaIndex through Composio's Tool Router MCP layer. Key takeaways:
  • Tool Router dynamically exposes Google cloud vision tools through an MCP endpoint
  • LlamaIndex's ReActAgent handles reasoning and orchestration; Composio handles integrations
  • The agent becomes more capable without increasing prompt size
  • Async Python provides clean, efficient execution of agent workflows
You can easily extend this to other toolkits like Gmail, Notion, Stripe, GitHub, and more by adding them to the toolkits parameter.
TOOLS

Supported Tools

Every Google cloud vision action and event your agent gets out of the box.

Annotate Files with Vision API

Tool to perform image detection and annotation for batch files in Google Cloud Vision.

Async Batch Annotate Files

Tool to run asynchronous image detection and annotation for a list of generic files (PDF, TIFF, GIF).

Annotate Images

Run image detection and annotation for a batch of images using Google Cloud Vision API.

Annotate Images Async Batch

Tool to run asynchronous image detection and annotation for a batch of images.

Annotate Location Images

Tool to run image detection and annotation for a batch of images scoped to a specific project and location.

Create Vision Product

Creates a new Product resource in Google Cloud Vision Product Search.

Create Product Set

Creates a new ProductSet resource in Google Cloud Vision Product Search.

Create ReferenceImage

Tool to create a ReferenceImage under a product.

Delete Product

Permanently deletes a Product and its associated reference images from Google Cloud Vision API.

Get Product

Tool to get information associated with a Product.

Get Product Set

Tool to get a ProductSet.

Import Product Sets

Asynchronously imports product sets and reference images from a CSV file stored in Google Cloud Storage.

List Vision AI IndexEndpoints

Lists IndexEndpoints in Vertex AI Vision for a given project and location.

List Locations

Tool to list available Vision AI service locations for a project.

List Vision API Operations

Tool to list operations that match the specified filter.

Purge Products

Tool to asynchronously delete products in a ProductSet or orphan products.

Update Product

Tool to update a Product's mutable fields: displayName, description, and productLabels.

Update Product Set

Tool to update a ProductSet resource.

Add Product to ProductSet

Add a Product to a ProductSet in Google Cloud Vision Product Search.

Cancel Vision Operation

Starts asynchronous cancellation of a long-running Vision API operation.

Delete Vision API Operation

Tool to delete a long-running Vision API operation.

Delete Product Set

Tool to permanently delete a ProductSet.

Delete Reference Image

Permanently removes a reference image from a product in Google Cloud Vision Product Search.

Get Vision API Operation

Retrieves the latest state of a long-running Vision API operation.

Get Reference Image

Tool to get information associated with a ReferenceImage.

List Products in ProductSet

Tool to list Products in a specified ProductSet.

List Projects

List Google Cloud projects accessible to the authenticated user via Cloud Resource Manager API.

List Reference Images

Tool to list reference images for a product.

Remove Product from ProductSet

Removes a Product from a specified ProductSet in Google Cloud Vision API.

FAQ

Frequently asked questions

With a standalone Google cloud vision MCP server, the agents and LLMs can only access a fixed set of Google cloud vision tools tied to that server. However, with the Composio Tool Router, agents can dynamically load tools from Google cloud vision and many other apps based on the task at hand, all through a single MCP endpoint.

Yes, you can. LlamaIndex fully supports MCP integration. You get structured tool calling, message history handling, and model orchestration while Tool Router takes care of discovering and serving the right Google cloud vision tools.

Yes, absolutely. You can configure which Google cloud vision scopes and actions are allowed when connecting your account to Composio. You can also bring your own OAuth credentials or API configuration so you keep full control over what the agent can do.

All sensitive data such as tokens, keys, and configuration is fully encrypted at rest and in transit. Composio is SOC 2 Type 2 compliant and follows strict security practices so your Google cloud vision data and credentials are handled as safely as possible.

Start with Google cloud vision.It takes 30 seconds.

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

Start building