Lindorm MCP Server

This repository is an example of how to create a MCP server for Lindorm, a multi-model NoSQL database.

Usage

Configuration on lindorm

To utilize this MCP server, follow these steps:

1. Purchase the Lindorm wide-table engine, search-engine, vector-engine, and AI-engine on Alibaba Cloud.
2. Deploy a text-embedding model by following the official instructions.
3. Create your index (knowledgebase) and import your data using the deployed embedding model.

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 with your specific configuration:

• LINDORM_INSTANCE_ID: Your Lindorm instance ID
• USING_VPC_NETWORK: Set to true if running on VPC network, otherwise false
• USERNAME: Your Lindorm account username
• PASSWORD: Your Lindorm account password
• TEXT_EMBEDDING_MODEL: The name of your deployed text-embedding model
• TABLE_DATABASE: The database for SQL operations Note: This configuration assumes all engines share the same username and password.

Running the MCP Server

You should install uv. Directly start the mcp server.

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 the .env file under /path/to/alibabacloud-lindorm-mcp-server/
3. Copy the MCP configuration from .vscode/mcp.json to cline_mcp_settings.json, replacing paths and variables as needed.
4. Start the MCP server through the Cline extension.

Components

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

Available Tools

• lindorm_retrieve_from_index: Retrieve from an existing indexes(or knowledgebase) using both full-text search and vector search, and return the aggregated results
  • Parameters
    • index_name: the index name, or known as knowledgebase name
    • query: the query that you want to search in knowledgebase
    • content_field: the text field that store the content text. You can get it from the index structure by lindorm_get_index_mappings tool
    • vector_field: the vector field that store the vector index. You can get it from the index structure by lindorm_get_index_mappings tool
    • top_k: the result number that you want to return
• lindorm_get_index_fields: Get the fields info of the indexes(or knowledgebase), especially get the vector stored field and content stored field.
  • Parameters:
    • index_name: the index name, or known as knowledgebase name
• lindorm_list_all_index: List all the indexes(or knowledgebase) you have.
• lindorm_execute_sql: Execute SQL query on Lindorm database.
  • Parameters
    • query: The SQL query to execute which start with select
• lindorm_show_tables: Get all tables in the Lindorm database
• lindorm_describe_table: Get tables schema in the Lindorm database
  • Parameters
    • table_name: the table name