Alibaba Cloud Lindorm

🚀 Lindorm MCP Server

This repository serves as an example of creating an MCP server for Lindorm, a multi-model NoSQL database.

🚀 Quick Start

Configuration on Lindorm

To utilize this MCP server, follow these steps:

1. Purchase the wide - table engine, search engine, vector engine, and AI engine of Lindorm on Alibaba Cloud.
2. Deploy a text embedding model according to the official guide.
3. Create an index (knowledge base) using the deployed embedding model and import data.

Environment Setup

1. Clone this repository and navigate to the project directory.
2. Create your environment file:

cp .env.example .env

3. Edit the .env file according to your specific needs:

• LINDORM_INSTANCE_ID: Your Lindorm instance ID
• USING_VPC_NETWORK: Set to true if running on a VPC network, otherwise false
• USERNAME: Your Lindorm account username
• PASSWORD: Your Lindorm account password
• TEXT_EMBEDDING_MODEL: The name of the deployed text embedding model
• TABLE_DATABASE: The database for executing SQL operations

Note: This configuration assumes that all engines share the same username and password.

Running the MCP Server

You need to install uv. Start the MCP server directly.

cd /path/to/alibabacloud-lindorm-mcp-server/

uv pip install .

uv run python -m src.lindorm_mcp_server.server

Visual Studio Code

1. Install the Cline extension.
2. Create a .env file under /path/to/alibabacloud-lindorm-mcp-server/.
3. Copy the MCP configuration from .vscode/mcp.json to cline_mcp_settings.json, and replace the paths and variables as needed.
4. Start the MCP server using the Cline extension.

✨ Features

Components

• LindormVectorSearchClient: Perform full - text and vector searches on the search and vector engines.
• LindormWideTableClient: Execute SQL operations on Lindorm wide tables.

Available Tools

• lindorm_retrieve_from_index: Perform full - text and vector searches from an existing index (or knowledge base) and return aggregated results.
  • Parameters:
    • index_name: The name of the index, i.e., the knowledge base name.
    • query: The query content to search in the knowledge base.
    • content_field: The field storing the content text. You can obtain it from the index structure using the lindorm_get_index_mappings tool.
    • vector_field: The field storing the vector index. You can obtain it from the index structure using the lindorm_get_index_mappings tool.
    • top_k: The number of results to return.
• lindorm_get_index_fields: Get the field information of an index (or knowledge base), especially the fields storing vectors and content.
  • Parameters:
    • index_name: The name of the index, i.e., the knowledge base name.
• lindorm_list_all_index: List all indexes (or knowledge bases).
• lindorm_execute_sql: Execute an SQL query on the Lindorm database.
  • Parameters:
    • query: An SQL query starting with select.
• lindorm_show_tables: Get all tables in the Lindorm database.
• lindorm_describe_table: Get the schema information of a table in the Lindorm database.
  • Parameters:
    • table_name: The name of the table.