cisco-mcp
Enables LLMs to run read-only show commands on Cisco IOS/IOS-XE and Nexus switches over SSH, with a two-account model for privilege separation and a fail-closed allowlist for safety.
README
cisco-mcp
A read-only MCP server for Cisco IOS / IOS-XE
and Nexus (NX-OS) switches. It lets an LLM run show commands over SSH — and
nothing else.
The two-account model
To run show running-config on IOS you need privilege 15, but you don't want a
priv-15 account doing everything. So the server uses two accounts:
| Account | Used for |
|---|---|
Read-only (CISCO_RO_*) |
Every normal tool call |
Privilege-15 (CISCO_PRIV15_*) |
Only commands that need priv 15 on IOS/IOS-XE (show running-config, show startup-config, show tech-support, …) |
Platform-aware: privilege levels are an IOS/IOS-XE concept. Nexus uses RBAC —
its read-only network-operator role can already read the running config — so on
NX-OS devices the server always uses the read-only account and never escalates.
Safety
Every command passes through a fail-closed allowlist (allowlist.py) before it
runs:
- must be a
showcommand (abbreviations likesh runincluded); - config mode,
write/erase/reload/copy/clear/debug, command chaining (;, newlines), and pipe-to-write (| redirect,| tee,| append) are rejected.
The LLM never decides what's safe — the server enforces it mechanically, which also contains prompt-injection arriving through arguments or command output.
Setup
# 1. install (uv recommended)
uv sync # or: pip install -e .
# 2. credentials
cp .env.example .env # fill in the two accounts
# 3. inventory
cp devices.example.yaml devices.yaml # list your switches + platform
# 4. run tests
uv run pytest
Register with an MCP client
stdio transport, e.g. in a client config:
{
"mcpServers": {
"cisco": {
"command": "uv",
"args": ["run", "cisco-mcp"],
"cwd": "/path/to/cisco-mcp"
}
}
}
Tools
list_devices— inventory with platform + notesget_version,get_interfaces,get_interface_status,get_vlans,get_cdp_neighborsget_running_config— priv-15 account on IOS, read-only on NX-OSrun_show_command(device, command)— any allowlistedshow, with the same gate
Layout
src/cisco_mcp/
server.py MCP tools (FastMCP, stdio)
allowlist.py safety gate + priv-15 policy <- security core
connection.py account selection + Netmiko SSH
credentials.py two account profiles from env
inventory.py devices.yaml loader
platforms.py IOS vs NX-OS behavior
tests/ allowlist + account-selection tests (no network)
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.