alibabacloud-lindorm

This MCP server provides a Lindorm integration, exposing tools to perform both full-text and vector searches against Lindorm indexes (knowledge bases) and to run SQL operations on Lindorm databases. The server exposes a set of tools including lindorm_retrieve_from_index for querying knowledgebases with aggregated results from text and vector indices, lindorm_get_index_fields to inspect index metadata (notably the content and vector fields), lindorm_list_all_index to enumerate all knowledgebases, lindorm_execute_sql to run SQL queries against Lindorm, lindorm_show_tables to list tables, and lindorm_describe_table to retrieve a table's schema. You can interact with these tools via the MCP runtime once the server is started with uv, using your Lindorm credentials and the configured embedding model. This enables you to build retrieval-augmented workflows where natural language queries are translated into SQL or vector/search operations against your Lindorm data.

Prerequisites:\n- Python 3.8+ installed on your system\n- uv (Uvicorn-like tooling for MCP) available via pip or your environment runner\n- Access to a Lindorm instance with the necessary engines (wide-table, search, vector, and AI engines) as described in the Lindorm setup.\n\nInstallation steps:\n1) Clone the repository: git clone https://github.com/your-org/aliyun-alibabacloud-lindorm-mcp-server.git\n2) Navigate to the project directory: cd aliyun-alibabacloud-lindorm-mcp-server\n3) Create and configure environment variables:\n - cp .env.example .env\n - Edit .env to include LINDORM_INSTANCE_ID, USING_VPC_NETWORK, USERNAME, PASSWORD, TEXT_EMBEDDING_MODEL, TABLE_DATABASE.\n4) Install the MCP runtime (uv):\n uv pip install .\n5) Run the MCP server:\n uv run python -m src.lindorm_mcp_server.server\n\nNotes:\n- Ensure your Lindorm credentials are kept secure.\n- If you run behind a VPC, set USING_VPC_NETWORK=true and ensure network access to Lindorm.\n- The embedding model name should match what you deployed in Lindorm.

Environment variables: LINDORM_INSTANCE_ID, USING_VPC_NETWORK, USERNAME, PASSWORD, TEXT_EMBEDDING_MODEL, TABLE_DATABASE are required for connecting to Lindorm and performing SQL and vector operations. If you encounter connection issues, verify network access to Lindorm, and ensure the embedding model name matches the deployed model. The available tools support both textual and vector search workflows; for example, lindorm_retrieve_from_index combines full-text and vector results, while lindorm_execute_sql runs standard SQL against Lindorm databases. When debugging, check the .env file and confirm that uv is correctly invoking the server module path (src.lindorm_mcp_server.server).