MCP AI Monitor Server

MCP_AI_Monitor is a complete system monitoring solution that uses unsupervised machine learning to detect abnormal resource usage in real time. It gives you visibility into CPU, RAM, and network metrics, adaptive anomaly scoring, and fast, actionable reports across multiple channels, including Discord. It’s designed to help you spot unusual behavior early, reduce false positives, and maintain stable system performance.

How to use

You will interact with MCP_AI_Monitor through a small CLI client. Start by collecting data, train the anomaly detection model, and then run real-time monitoring to see anomalies as they occur. You can also generate usage statistics, send reports to Discord, and run a full automated sequence that covers collection, training, and monitoring.

Recommended usage workflow:

• Collect initial data to establish a baseline
• Train the Isolation Forest model on the collected data
• Run the real-time monitor to detect anomalies as your system evolves
• Optionally generate graphs and reports, and push insights to Discord for team visibility

If you want to run a complete end-to-end process in one go, you can execute the full flow and have a report sent to Discord. This combines collection, training, monitoring, and reporting in a single command.

How to install

# Prerequisites
- Python 3.8+
- pip (comes with Python 3.8+)

# Clone the project
git clone https://github.com/MedusaSH/MCP_AI_Monitor.git
cd MCP_AI_Monitor

# Install dependencies
pip install -r requirements.txt

# Optional: configure Discord webhooks for reports (see configuration notes below)

Notes:

• The project relies on Python packages such as psutil, scikit-learn, pandas, matplotlib, colorama, and discord-webhook. Ensure you have network access to install dependencies. You may need to use a virtual environment for isolation.

Additional content

Configuration tips and workflow notes follow. The system supports sending detailed hardware and network reports to Discord via webhooks. You can tailor webhook URLs to separate channels for hardware metrics and network data.

Usage examples include:

• Collect data: python mcp.py collect
• Train model: python mcp.py train
• Start real-time monitoring: python mcp.py monitor
• Generate statistics: python mcp.py stats
• Send Discord reports: python mcp.py discord
• Analyze network and report: python mcp.py network
• Run the full sequence: python mcp.py all --duration 120 --report

Tools and endpoints

The available command sets let you perform specific actions and obtain targeted reports. Key tools include data collection, model training, live anomaly monitoring, and reporting for both hardware and network insights.

Available tools

collect

Collects system metrics (CPU, RAM, etc.) at regular intervals and stores them for analysis.

train

Preprocesses gathered data and trains an Isolation Forest model to detect anomalies.

monitor

Runs real-time monitoring using the trained model and adapts thresholds over time.

stats

Generates graphs and statistics about resource usage for review.

discord

Sends detailed reports to Discord via configurable webhooks.

network

Analyzes network activity and produces a dedicated report.

all

Executes the full sequence: collect, train, monitor, and optionally report.