MCP ExplorerExplorer

Mcp Agent Ot

@pkbythebay29on 9 months ago
4 MIT
FreeCommunity
AI Systems
Multi-Agent Context Processor for Operational Technology (OT) data.

Overview

What is Mcp Agent Ot

mcp-agent-ot is a Multi-Agent Context Processor designed for Operational Technology (OT) data, enabling the connection to various industrial data sources and processing them efficiently.

Use cases

Use cases include real-time monitoring of industrial systems, anomaly detection in operational data, context-aware Q&A for operators, and integration with edge computing solutions.

How to use

To use mcp-agent-ot, clone the repository, install the required dependencies, start the system, and access the dashboard via http://localhost:8000. Configure data sources in the data_sources.yml file.

Key features

Key features include support for MQTT, OPC UA, and Modbus sources, asynchronous agent operation, a vector database per zone using FAISS, Q&A capabilities with MiniLM, a FastAPI dashboard for monitoring, and a modular design for extensibility.

Where to use

mcp-agent-ot can be used in various fields related to industrial automation, IoT, and data processing in operational technology environments.

Content

🧠 mcp-agent-ot

Multi-Agent Context Processor for Operational Technology (OT) data.

Connects to industrial data sources (MQTT, OPC UA, Modbus), filters signals, embeds context into vector stores, and enables lightweight LLM-based Q&A — all using modular agents and Redis pub/sub.


🔧 Features

  • 🛰️ Plug in MQTT, OPC UA, or Modbus sources
  • 🔁 Agents run asynchronously (filtering, vectorizing, answering)
  • 📦 Vector DB per zone using FAISS
  • 🧠 Q&A using MiniLM (via sentence-transformers)
  • 🌐 FastAPI dashboard to monitor messages and triggers
  • 🧩 Modular, extensible, and modern Python packaging (pyproject.toml)

🚀 Quick Start

1. Clone the Repo

2. Install Dependencies

3. Start the system

4. Start the dashboard

🔌 Configuring Data Sources

Edit mcpai/configs/data_sources.yml:

🧱 Architecture

ingest_agent.py – connects to all OT sources dynamically

filter_agent.py – identifies meaningful anomalies (e.g., "high temp")

vector_agent.py – builds FAISS DB per zone

llm_agent.py – answers based on vector context

coordinator.py – starts everything

dashboard.py – FastAPI UI

redis_pubsub.py – pub/sub messaging layer

📄 License

MIT

🤝 Contributing

PRs and issues welcome. Designed to be extensible for edge computing and OT AI use cases.

Tools

No tools

Comments

Recommend MCP Servers

View All MCP Servers