IoTDB Time Series

Clone the official Apache IoTDB MCP server repository, create a virtual environment using uv, and install all required dependencies.

git clone https://github.com/apache/iotdb-mcp-server.git
cd iotdb-mcp-server
uv venv
source .venv/bin/activate
uv sync

Start your Apache IoTDB server if it is not already running. The MCP server will connect to it using the host, port, and credentials you configure. By default it expects IoTDB on 127.0.0.1:6667.

# Start IoTDB (from your IoTDB installation directory)
./sbin/start-standalone.sh

Set the environment variables that control the IoTDB connection. All variables have defaults but you should override them to match your setup, especially IOTDB_HOST, IOTDB_USER, and IOTDB_PASSWORD.

export IOTDB_HOST=127.0.0.1
export IOTDB_PORT=6667
export IOTDB_USER=root
export IOTDB_PASSWORD=root
export IOTDB_DATABASE=test
export IOTDB_SQL_DIALECT=table  # 'table' or 'tree'
export IOTDB_EXPORT_PATH=/tmp

Edit claude_desktop_config.json to register the IoTDB MCP server. You must provide the absolute path to the uv executable and the cloned repository location.

{
  "mcpServers": {
    "iotdb": {
      "command": "/usr/local/bin/uv",
      "args": ["run", "--directory", "/path/to/iotdb-mcp-server", "mcp"],
      "env": {
        "IOTDB_HOST": "127.0.0.1",
        "IOTDB_PORT": "6667",
        "IOTDB_USER": "root",
        "IOTDB_PASSWORD": "root",
        "IOTDB_DATABASE": "test",
        "IOTDB_SQL_DIALECT": "table",
        "IOTDB_EXPORT_PATH": "/tmp"
      }
    }
  }
}

With the server running, use your AI client to explore schemas and run time series queries. In table model mode the tools are `read_query`, `list_tables`, `describe_table`, and `export_table_query`. In tree model mode the tools are `select_query`, `metadata_query`, and `export_query`.