Network MCP Docker Suite: Seven MCP Servers for AI-Driven Network Operations
3 min read
Seven containerized MCP servers that bring AI assistants like Cursor, LibreChat, and Claude Desktop directly into your network operations – and why I built them.
This article is Part 2 of a 3-part series:
- Part 1: When Networks Understand What You Want – The MCP Protocol Revolution
- Part 2 (you are here): Network MCP Docker Suite – Seven MCP Servers for AI-Driven Network Operations
- Part 3: From NetDevOps to AgenticOps – Real-World Use Cases with the Network MCP Docker Suite
Real-World Problem Solving
Before explaining what the Network MCP Docker Suite is, let’s start with a typical problem many network-operations teams face.
Traditional Troubleshooting Workflow
Picture this scenario: a NOC engineer receives an alert that a network device is unreachable from Cisco Catalyst Center. The traditional troubleshooting workflow often looks like this:
- Log into Catalyst Center
- Navigate to the device
- Check status and metrics
- Open an SSH client
- Connect to the device and run diagnostic commands
- Check neighbors and interfaces
- Cross-reference with documentation
- Identify and fix the issue
Time consumed: 15–30 minutes or more, especially if data sources are scattered.
AI-Powered Troubleshooting with MCP
Now imagine the same engineer simply asking an AI assistant:
Engineer: “Check why wlsn-access-1.dna.its-best.ch is unreachable from Cisco Catalyst Center.”
AI-driven troubleshooting in action: natural-language queries automatically investigate and resolve issues through multiple MCP servers.
The assistant, connected to multiple MCP servers, can:
- Query Catalyst Center for device status and known issues
- Connect to the device via SSH using the IOS XE MCP Server
- Verify connectivity and neighbors
- Correlate data from both systems using AI reasoning
- Identify the root cause (e.g., inventory IP mismatch)
- Suggest corrective steps
Time consumed: often under two minutes.
This is the kind of workflow I wanted to enable in a repeatable, easily deployable way. That’s where the Network MCP Docker Suite comes in.
What is the Network MCP Docker Suite?
In Part 1, I introduced the Model Context Protocol (MCP) – an open standard that lets AI agents interact with tools and data sources consistently.
To bring these concepts to life, I built a community open-source project — the Network MCP Docker Suite — showing how AI agents can talk to network APIs using MCP.
The Network MCP Docker Suite is an open-source, Docker-based collection of seven MCP servers that enable AI-driven network operations through integration with Cisco platforms, NetBox, ThousandEyes, and Splunk.
Think of it as a universal translator that lets AI assistants like Cursor, LibreChat, and Claude Desktop communicate directly with your infrastructure using natural language – all powered by MCP.
The Seven Network MCP Servers
- ️ Meraki MCP Server: Cloud network management via Meraki Dashboard API
- NetBox MCP Server: DCIM/IPAM documentation and management
- Catalyst Center MCP Server: Enterprise network management and assurance
- IOS XE MCP Server: Direct device access via SSH
- ThousandEyes MCP Server: Performance monitoring and path visualization
- ISE MCP Server: Identity and access control operations
- Splunk MCP Server: Log analysis and operational intelligence
Each server exposes MCP tools that the AI assistant can call. Because they all speak MCP, an AI assistant can orchestrate them together in a single flow – just like in the example above.
Getting Started with the Network MCP Docker Suite
If you want to try this in your own lab, the Network MCP Docker Suite is available on GitHub with setup guides and examples.
GitHub Repository: github.com/pamosima/network-mcp-docker-suite
The repository includes:
- Pre-built Docker containers for all seven servers
- Setup guides and configuration examples
- Flexible deployment profiles (all, Cisco, monitoring, security)
- Environment templates for quick start
- Integration examples for Cursor, LibreChat, and Claude Desktop
- Troubleshooting tips and documentation
The Network MCP Docker Suite was also featured in a recent live stream on the Cisco DevNet YouTube channel, where we explored how AI agents interact with network APIs through MCP servers.
▶ Watch the Cisco DevNet live session on YouTube
What Comes Next?
The Network MCP Docker Suite focuses on making MCP tangible for network engineers and operations teams. In Part 3, I explore how these capabilities connect to the broader vision of AgenticOps – including Cisco’s AI Canvas and Deep Network Model.
Continue reading the series:
- Back to Part 1: When Networks Understand What You Want – The MCP Protocol Revolution
- Next – Part 3: From NetDevOps to AgenticOps – Real-World Use Cases with the Network MCP Docker Suite
Note: The Network MCP Docker Suite is a personal open-source project created for demonstration purposes. It is not an official Cisco product. The examples and code are provided “as is” to illustrate how AI agents can interact with network APIs through the Model Context Protocol (MCP).
Authors
