EdgeLake

You can run an EdgeLake MCP Server to access and query EdgeLake data across nodes using a lightweight, stateless MCP interface. This server lets AI assistants discover resources, inspect table schemas, and execute powerful SQL queries with filtering, joins, aggregations, and extended metadata.

How to use

To use this MCP server, you run the local MCP process and connect your MCP client (such as Claude Desktop) to it. The server exposes commands to discover databases and tables, read table schemas, run complex queries, and check node health. Your client can then request resource lists, schemas, and query results across EdgeLake databases and tables.

How to install

Prerequisites and first steps ensure you can run the MCP server and connect your client.

# 1) Install Python 3.10 or higher (if not already installed)
# 2) Create a project directory and install dependencies

python -m venv venv-edgelake-mcp
source venv-edgelake-mcp/bin/activate

# 3) Install required Python packages
pip install -r requirements.txt

# 4) Prepare environment variables (see configuration section for details)
# 5) Run the MCP server locally
python server.py

Configuration and environment

Configure how the server talks to the EdgeLake node using environment variables. The following settings are used by default and can be overridden in a .env file or your shell environment.

EDGELAKE_HOST=127.0.0.1
EDGELAKE_PORT=32049
EDGELAKE_TIMEOUT=20
EDGELAKE_MAX_WORKERS=10
LOG_LEVEL=INFO

Running the server

Start the MCP server as a standard Python process. It runs as a subprocess using a simple stdio transport. Use this command exactly as shown.

python server.py

MCP client configuration (how to connect)

Add your EdgeLake MCP server connection to your client configuration. The example below demonstrates how to configure a local, stdio-based server named edgelake within your client app. Adapt the path to your server accordingly.

{
  "mcpServers": {
    "edgelake": {
      "command": "python",
      "args": ["server.py"],
      "env": {
        "EDGELAKE_HOST": "192.168.1.106",
        "EDGELAKE_PORT": "32049"
      }
    }
  }
}

Available tools

resources/list

Lists all available databases and tables across EdgeLake nodes, enabling discovery of what data you can access.

resources/read

Reads a specific table schema, returning column definitions and metadata for the target database and table.

query

Executes an SQL query against EdgeLake with advanced filtering, grouping, ordering, JOINs, and optional extended metadata fields.

node_status

Returns health and status information about the EdgeLake MCP server and its connection to EdgeLake nodes.

list_databases

Lists all databases available on the connected EdgeLake node for discovery before querying.

list_tables

Lists all tables within a specified database to help you understand available data.

get_schema

Retrieves the schema (columns) for a given table to guide query construction.

server_info

Provides the MCP server version and current configuration information.