Lumenore Analytics

You run a Model Context Protocol (MCP) server that connects AI assistants to Lumenore’s analytics capabilities. It enables natural language queries to be transformed into data insights, supports real-time streaming for large datasets, and runs in memory without persisting user data, while enforcing Lumenore’s privacy policies.

How to use

Connect your MCP client to the server using the provided HTTP endpoint. Expect to send natural language queries that are translated into data actions such as fetching datasets, calculating trends, forecasting, detecting anomalies, and exploring correlations. Start with a dataset discovery step, then issue focused queries to obtain insights, and finally combine tools to build a complete analysis workflow. Use streaming transport to receive responses as data is generated.

How to install

Prerequisites: Python 3.13+, network access to reach the Lumenore backend, and valid client credentials.

Clone the project, install dependencies, and prepare configuration files with your credentials.

git clone https://github.com/Lumenore-Platform/lumenore-mcp.git
cd lumenore-mcp
pip install -r requirements.txt
cp .env.example .env
# Edit .env with your client credentials (LUMENORE_CLIENT_ID and LUMENORE_SECRET)
python main.py

Configuration

Create a .env file in the project root and populate the required credentials. You can use .env.example as a template.

Using client credentials, set the following values in the environment file:

LUMENORE_CLIENT_ID="your_client_id_here"
LUMENORE_SECRET="your_secret_here"
SERVER_URL="https://preview.lumenore.com"

Security & Privacy

Data is processed in-memory and discarded after generating a response. No query data is stored by the MCP server, and all backend communications use HTTPS. Client credentials are kept out of logs and responses, and each deployment should rotate credentials regularly.

Troubleshooting

If the server does not respond as expected, perform quick checks: ensure the server is running, verify the environment variables are loaded, and confirm the backend URL is reachable. Test the endpoint quickly with a lightweight request and review startup logs for authentication or transport errors.

Usage examples

After starting the server, you can use the tools to analyze data. Begin by discovering datasets, then convert natural language queries into data actions, and finally apply analytics tools such as trends, forecasts, and Pareto analysis to derive actionable insights.

Available tools

get_dataset_metadata

Lists available datasets and schema IDs to help you discover what data you can analyze next.

nlq_to_data

Converts natural language questions into structured data queries using a specified schema.

get_trend_data

Identifies temporal patterns and trends from the selected dataset to show how values change over time.

get_prediction_data

Generates forecasts and predictions for future periods based on historical data.

get_outlier_data

Detects anomalies and unusual patterns that stand out from normal behavior.

get_correlation_data

Analyzes relationships between variables to explain dependencies and drivers.

get_change_data

Detects shifts and transitions in data patterns over time.

get_pareto_data

Performs 80/20 analysis to identify the most impactful contributors.