EVE-NG MCP Server
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@EVE-NG MCP Server Create a lab with two routers connected via Ethernet."
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
EVE-NG MCP Server
An MCP (Model Context Protocol) server that gives Claude and other LLM agents programmatic access to EVE-NG network labs.
Create, configure, and manage virtual network topologies through natural language — no manual clicking required.
You: "Create a lab with two Arista switches running OSPF, connected to each other"
Claude: ✓ Created lab "OSPF Demo"
✓ Added vEOS-1 (Arista vEOS 4.28)
✓ Added vEOS-2 (Arista vEOS 4.28)
✓ Connected e0/0 ↔ e0/0
✓ Pushed OSPF configs
✓ Started all nodes — lab is runningQuick Start
1. Install
pip install eveng-mcp-server2. Configure
Set your EVE-NG connection details:
export EVENG_HOST=http://your-eve-ng-server
export EVENG_USERNAME=admin
export EVENG_PASSWORD=eve3. Add to Claude
Claude Desktop — add to ~/Library/Application Support/Claude/claude_desktop_config.json:
{
"mcpServers": {
"eveng": {
"command": "eveng-mcp-server",
"env": {
"EVENG_HOST": "http://your-eve-ng-server",
"EVENG_USERNAME": "admin",
"EVENG_PASSWORD": "eve"
}
}
}
}Claude Code — add to .claude/settings.json:
{
"mcpServers": {
"eveng": {
"command": "eveng-mcp-server",
"env": {
"EVENG_HOST": "http://your-eve-ng-server",
"EVENG_USERNAME": "admin",
"EVENG_PASSWORD": "eve"
}
}
}
}4. Use
Ask Claude to build you a network lab. That's it.
Related MCP server: Packet Tracer MCP
How It Works
┌─────────────┐ MCP Protocol ┌──────────────────┐ REST API ┌─────────────┐
│ Claude / │◄──────────────────────►│ EVE-NG MCP │◄───────────────►│ EVE-NG │
│ LLM Agent │ (stdio transport) │ Server │ (HTTP + cookies)│ Server │
└─────────────┘ └──────────────────┘ └─────────────┘
│
14 MCP Tools:
create_lab, add_node,
connect_nodes, push_config,
start_all, etc.The MCP server translates natural language requests from Claude into EVE-NG REST API calls. It handles authentication (session cookies), lab lifecycle, node management, configuration push, and network connectivity.
Available Tools
Lab Management
Tool | Description |
| List all labs on the server |
| Get lab details — nodes, networks, links, status |
| Create a new empty lab |
| Delete a lab and all its resources |
Node Operations
Tool | Description |
| Browse all installed QEMU and Docker images |
| Add a node to a lab — specify image, name, RAM, CPU, interface count |
| Start a single node |
| Stop a single node |
| Start all nodes in a lab |
| Stop all nodes in a lab |
| Get a node's current status and configuration |
Configuration
Tool | Description |
| Push startup configuration to a node (must be stopped) |
| Retrieve a node's current startup configuration |
Networking
Tool | Description |
| Connect two node interfaces — creates a shared network and wires both endpoints |
Usage Examples
Build a Multi-Vendor Lab
"Create a lab called 'Data Center Fabric' with 2 VOSS spine switches and 4 EXOS leaf switches. Connect each leaf to both spines."
Claude will use list_images to find your installed Extreme images, create the lab, add 6 nodes, and wire the spine-leaf fabric.
Push Configs Before Boot
"Push OSPF configuration to all routers in the lab, then start everything."
Claude will call push_config on each node with vendor-appropriate configuration, then start_all to boot the lab.
Inspect Lab State
"Show me what's running in the BGP lab — which nodes are up, which are still booting?"
Claude calls get_lab to retrieve the full topology with per-node status.
Tear Down and Rebuild
"Destroy the current lab and build a new one with 3 Cisco IOSv routers in a triangle topology."
Claude calls delete_lab, then create_lab, add_node x3, and connect_nodes x3.
Common Use Cases
Certification Lab Prep (CCNA/CCNP/CCIE)
"Build me a lab for practicing OSPF multi-area with 5 routers — one ABR connecting area 0 and area 1, with stub and NSSA areas."
The MCP server handles the tedious setup: creating the lab, adding nodes, wiring interfaces, and pushing base configs so you can focus on the protocol practice.
Pre-Change Validation
"I need to test adding a new VLAN to our spine-leaf fabric before doing it in production. Build a lab that mirrors our 2-spine 4-leaf topology with VOSS switches."
Build a lab topology matching production, test your change, verify traffic flow — all through conversation with Claude.
Vendor Comparison / PoC
"Create two identical 3-router triangle topologies — one with Arista vEOS and one with Cisco IOSv. Push BGP configs to both so I can compare convergence behavior."
Side-by-side vendor comparison without manually building two separate labs.
Training and Demos
"Set up a classroom lab with 5 separate student topologies, each with 2 routers and 1 switch, all pre-configured with basic connectivity."
Quickly spin up multiple isolated lab instances for training sessions.
Network Automation Development
"Create a 4-router lab and start all nodes. I need to test my Ansible playbooks against real network devices."
Use the MCP server to provision labs for testing Ansible, Nornir, Terraform, or any network automation tooling.
Troubleshooting Practice
"Build a lab with intentional misconfigurations — wrong OSPF area IDs, mismatched BGP AS numbers, or broken VLAN trunks — and don't tell me what's wrong."
AI-generated fault injection for network troubleshooting practice.
Prerequisites
Python 3.10+
EVE-NG server (Community or Professional) with API access enabled
The server must be reachable from where the MCP server runs
Default EVE-NG API port is 80 (HTTP) or 443 (HTTPS)
Default credentials:
admin/eve
MCP-compatible client — Claude Desktop, Claude Code, or any MCP client
EVE-NG Network Access
The MCP server connects to EVE-NG's REST API over HTTP. Your EVE-NG server must be reachable:
Setup | EVENG_HOST Value |
Same machine |
|
Local network |
|
Via Tailscale |
|
Via SSH tunnel |
|
Configuration Reference
Variable | Default | Description |
|
| EVE-NG API base URL |
|
| API username |
|
| API password |
Troubleshooting
"Connection refused" or timeout
Verify EVE-NG is running and the API is accessible:
curl http://your-server/api/statusCheck firewall rules — EVE-NG API runs on port 80 by default
If using a jump host, set up an SSH tunnel first
"Authentication failed"
Default credentials are
admin/eveEVE-NG Community Edition limits to 2 admin accounts
Check if another session is active (EVE-NG allows one session per user)
"Image not found" when adding nodes
Run
list_imagesto see what's installedEVE-NG image folder names must follow exact naming conventions
QEMU images go in
/opt/unetlab/addons/qemu/
Node starts but no console access
Nodes need 30-120 seconds to fully boot (vendor dependent)
Use
get_node_statusto check if the node is still building (status=1) or running (status=2)
Development
# Clone and install
git clone https://github.com/axiom-works-ai/eveng-mcp-server.git
cd eveng-mcp-server
pip install -e ".[dev]"
# Run tests (69 tests, mocked — no EVE-NG server needed)
pytest
# Run with verbose output
pytest -v
# Type check
mypy src/eveng_mcp_server/ --ignore-missing-importsEVE-NG API Notes
Uses session cookies for authentication (not bearer tokens)
Lab paths are URL-encoded filesystem paths (e.g.,
/api/labs/My%20Lab.unl)Node status codes:
0= stopped,1= building/booting,2= runningPush startup configs while nodes are stopped — they apply at next boot
connect_nodesis a compound operation: creates a network, then attaches both interfaces
License
Author
Built by Axiom Works AI — AI-powered tools for network engineers.
Available Tools
14 toolsadd_nodeB
Add a node (virtual device) to a lab. Specify the template (image name), name, RAM, CPU, and number of interfaces. Use 'list_images' first to see available templates.
Inputs: lab_path (required), template (required), name, node_type, +10 more. Returns: result
connect_nodesB
Connect two nodes by creating a network and attaching their interfaces to it. Specify the node IDs and interface IDs to connect. Use 'get_lab' first to see available nodes and their interface IDs.
Inputs: lab_path (required), node1_id (required), node1_interface (required), node2_id (required), +3 more. Returns: result
create_labB
Create a new lab on the EVE-NG server. Specify a name, optional description, and folder path.
Inputs: name (required), description, path, author, +2 more. Returns: result
delete_labC
Delete a lab from the EVE-NG server. This is irreversible.
Inputs: lab_path (required). Returns: result
get_labB
Get detailed information about a lab including its nodes, networks, and topology. Returns lab metadata, all nodes with their status, and all networks.
Inputs: lab_path (required). Returns: result
get_node_configB
Get a node's current startup configuration text.
Inputs: lab_path (required), node_id (required). Returns: result
get_node_statusA
Get the current status of a node — whether it is running, stopped, or building. Also returns node details like RAM, CPU, template, and console URL.
Inputs: lab_path (required), node_id (required). Returns: result
list_imagesA
List all available QEMU/Docker node images (templates) installed on the EVE-NG server. Returns template names and descriptions.
Inputs: no named parameters. Returns: result
list_labsA
List all labs on the EVE-NG server. Returns lab names, filenames, and paths. Optionally specify a folder path to list labs in a subfolder.
Inputs: folder. Returns: result
push_configA
Push a startup configuration to a node. The node should be stopped before pushing config. The configuration is stored and applied on next boot.
Inputs: lab_path (required), node_id (required), config (required). Returns: result
start_allB
Start all nodes in a lab simultaneously.
Inputs: lab_path (required). Returns: result
start_nodeC
Start a single node in a lab. The node must already exist.
Inputs: lab_path (required), node_id (required). Returns: result
stop_allC
Stop all nodes in a lab simultaneously.
Inputs: lab_path (required). Returns: result
stop_nodeB
Stop a single node in a lab.
Inputs: lab_path (required), node_id (required). Returns: result
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 14 tools
Each tool targets a distinct operation (creating labs, adding nodes, connecting, starting/stopping, configuration), with no overlap or ambiguity.
All tool names follow a consistent verb_noun snake_case pattern (e.g., create_lab, get_lab, start_node), making them predictable.
14 tools appropriately cover lab, node, connectivity, and configuration management without being excessive or sparse.
Core workflows (create lab, add nodes, connect, configure, start/stop) are covered; missing operations like delete node or remove connection are minor gaps.
Maintenance
Related MCP Connectors
MCP server for AI agents to plan, verify, and deploy Cloudflare-native apps.
An MCP server that provides an API to LLMs to manage their JumpCloud resources.
MCP server unifying ERPs, CRMs, APIs and knowledge base for Claude, ChatGPT and Gemini.
MCP server that lets your AI assistant build and deploy apps, hosting each one at a live URL.
Related MCP Servers
- FlicenseAqualityDmaintenanceA Model Context Protocol server that enables LLMs to manage network topologies, labs, nodes, and configurations in the EVE-NG network emulation platform through a standardized interface.2615-
- AlicenseNot gradedqualityAmaintenanceAn MCP server that allows LLMs to create, configure, validate, and explain Cisco Packet Tracer network topologies. It provides a comprehensive suite of tools for generating deployment scripts, CLI configurations, and automated network troubleshooting.2,699 PyPI377MIT
- FlicenseAqualityFmaintenanceAn MCP server that integrates Nornir with NAPALM and Netmiko, enabling LLMs to orchestrate multi-vendor network infrastructure through natural language.52-
- AlicenseAqualityCmaintenanceMCP server for controlling PNETLab v6 network labs via natural language, enabling topology creation, node management, config push, console interaction, and link fault injection.241Apache 2.0