MCP Vertexai Search

🚀 MCP Server for Searching with Vertex AI

This is an MCP server designed for document search using Vertex AI. It enables users to leverage Vertex AI's capabilities to efficiently search through documents, providing a powerful solution for information retrieval.

🚀 Quick Start

✨ Features

• Enhanced Search with Grounding: Utilizes Gemini combined with Vertex AI's Grounding feature to search documents using private data from Vertex AI data warehouses. Grounding technology improves the quality of Gemini's responses, generating more accurate results based on your stored data.
• Flexible Deployment: Supports two ways to use the MCP server, either through Docker or by installing the Python package.
• Multiple Transport Protocols: Supports both SSE (Server-Sent Events) and stdio (standard input/output) for data transmission.

📦 Installation

There are two ways to use this MCP server. The first method is more convenient if you want to run it on Docker, as a Dockerfile is provided in the project.

Method 1: Clone the Repository

# Clone the repository
git clone git@github.com:ubie-oss/mcp-vertexai-search.git

# Create a virtual environment
uv venv
# Install dependencies
uv sync --all-extras

# View the running command
uv run mcp-vertexai-search

Method 2: Install the Python Package

The package is not yet published on PyPI, but we can install it through the repository. Since you need to use a configuration file exported from config.yml.template to run the MCP server, you must modify this template file.

# Install the package
pip install git+https://github.com/ubie-oss/mcp-vertexai-search.git

# View the help command
mcp-vertexai-search --help

💻 Usage Examples

Basic Usage

There are two ways to use this MCP server. You can choose the appropriate method based on your needs.

Advanced Usage

This implementation supports two transport protocols, SSE (Server-Sent Events) and stdio (standard input/output). You can select different transport methods by setting the --transport flag.

You can configure the MCP server using a YAML file. config.yml.template is a template for the configuration file. Please refer to Appendix A: Configuration File for detailed information and modify the configuration file as needed.

uv run mcp-vertexai-search serve \
    --config config.yml \
    --transport <stdio|sse>

📚 Documentation

Architecture

This solution uses Gemini in combination with Vertex AI's Grounding feature to search documents using private data from Vertex AI data warehouses. Grounding technology can improve the quality of Gemini's responses, enabling it to generate more accurate results based on the data stored in your Vertex AI data warehouse. We can achieve this by integrating one or more Vertex AI data warehouses into the MCP server.

For more detailed information about Grounding, please refer to the Vertex AI Grounding official documentation.

Development

Prerequisites

• uv
• Vertex AI data warehouse
  • Please refer to the official documentation for more information about the data warehouse.

Set Up the Local Environment

# Optional: Install uv
python -m pip install -r requirements.setup.txt

# Create a virtual environment
uv venv
uv sync --all-extras

Run the MCP Server

uv run mcp-vertexai-search serve \
    --config config.yml \
    --transport <stdio|sse>

Test Vertex AI Search

You can use the mcp-vertexai-search search command (without running the MCP server) to test the Vertex AI search function.

uv run mcp-vertexai-search search \
    --config config.yml \
    --query <your-query>

Appendix A: Configuration File

config.yml.template is a template for the configuration file.

Please fill in the above fields according to the actual situation to complete the configuration.