pyATS MCP Server

1# pyATS MCP Server
2[![Trust Score](https://archestra.ai/mcp-catalog/api/badge/quality/automateyournetwork/pyATS_MCP)](https://archestra.ai/mcp-catalog/automateyournetwork__pyats_mcp)
3 
4This project implements a Model Context Protocol (MCP) Server that wraps Cisco pyATS and Genie functionality. It enables structured, model-driven interaction with network devices over STDIO using the JSON-RPC 2.0 protocol.
5 
6🚨 This server does not use HTTP or SSE. All communication is done via STDIN/STDOUT (standard input/output), making it ideal for secure, embedded, containerized, or LangGraph-based tool integrations.
7 
8🔧 What It Does
9 
10Connects to Cisco IOS/NX-OS devices defined in a pyATS testbed
11 
12Supports safe execution of validated CLI commands (show, ping)
13 
14Allows controlled configuration changes
15 
16Returns structured (parsed) or raw output
17 
18Exposes a set of well-defined tools via tools/discover and tools/call
19 
20Operates entirely via STDIO for minimal surface area and maximum portability
21 
22🚀 Usage
23 
241. Set your testbed path
25 
26```bash
27 
28export PYATS_TESTBED_PATH=/absolute/path/to/testbed.yaml
29 
30```
31 
322. Run the server
33 
34Continuous STDIO Mode (default)
35 
36```bash
37 
38python3 pyats_mcp_server.py
39 
40```
41 
42Launches a long-running process that reads JSON-RPC requests from stdin and writes responses to stdout.
43 
44One-Shot Mode
45 
46``` bash
47 
48echo '{"jsonrpc": "2.0", "id": 1, "method": "tools/discover"}' | python3 pyats_mcp_server.py --oneshot
49 
50```
51Processes a single JSON-RPC request and exits.
52 
53📦 Docker Support
54 
55Build the container
56 
57```bash
58 
59docker build -t pyats-mcp-server .
60 
61```
62 
63Run the container (STDIO Mode)
64```bash
65docker run -i --rm \
66  -e PYATS_TESTBED_PATH=/app/testbed.yaml \
67  -v /your/testbed/folder:/app \
68  pyats-mcp-server
69```
70 
71🧠 Available MCP Tools
72 
73Tool	Description
74 
75run_show_command	Executes show commands safely with optional parsing
76 
77run_ping_command	Executes ping tests and returns parsed or raw results
78 
79apply_configuration	Applies safe configuration commands (multi-line supported)
80 
81learn_config	Fetches running config (show run brief)
82 
83learn_logging	Fetches system logs (show logging last 250)
84 
85All inputs are validated using Pydantic schemas for safety and consistency.
86 
87🤖 LangGraph Integration
88 
89Add the MCP server as a tool node in your LangGraph pipeline like so:
90 
91```python
92 
93("pyats-mcp", ["python3", "pyats_mcp_server.py", "--oneshot"], "tools/discover", "tools/call")
94 
95```
96 
97Name: pyats-mcp
98 
99Command: python3 pyats_mcp_server.py --oneshot
100 
101Discover Method: tools/discover
102 
103Call Method: tools/call
104 
105STDIO-based communication ensures tight integration with LangGraph’s tool invocation model without opening HTTP ports or exposing REST endpoints.
106 
107📜 Example Requests
108 
109Discover Tools
110 
111```json
112 
113{
114  "jsonrpc": "2.0",
115  "id": 1,
116  "method": "tools/discover"
117}
118 
119```
120 
121Run Show Command
122 
123``` json
124 
125{
126  "jsonrpc": "2.0",
127  "id": 2,
128  "method": "tools/call",
129  "params": {
130    "name": "run_show_command",
131    "arguments": {
132      "device_name": "router1",
133      "command": "show ip interface brief"
134    }
135  }
136}
137``` 
138🔒 Security Features
139 
140Input validation using Pydantic
141 
142Blocks unsafe commands like erase, reload, write
143 
144Prevents pipe/redirect abuse (e.g., | include, >, copy, etc.)
145 
146Gracefully handles parsing fallbacks and errors
147 
148📁 Project Structure
149 
150```graphql
151 
152.
153├── pyats_mcp_server.py     # MCP server with JSON-RPC and pyATS integration
154├── Dockerfile              # Docker container definition
155├── testbed.yaml            # pyATS testbed (user-provided)
156└── README.md               # This file
157 
158```
159 
160📥 MCP Server Config Example (pyATS MCP via Docker)
161 
162To run the pyATS MCP Server as a container with STDIO integration, configure your mcpServers like this:
163 
164``` json
165{
166  "mcpServers": {
167    "pyats": {
168      "command": "docker",
169      "args": [
170        "run",
171        "-i",
172        "--rm",
173        "-e",
174        "PYATS_TESTBED_PATH",
175        "-v",
176        "/absolute/path/to/testbed/folder:/app",
177        "pyats-mcp-server"
178      ],
179      "env": {
180        "PYATS_TESTBED_PATH": "/app/testbed.yaml"
181      }
182    }
183  }
184}
185 
186```
187 
188```json
189{
190  "servers": { 
191    "pyats": {
192      "type": "stdio",
193      "command": "python3",
194      "args": [
195        "-u",
196        "/Users/johncapobianco/pyATS_MCP/pyats_mcp_server.py"
197      ],
198      "env": {
199        "PYATS_TESTBED_PATH": "/Users/johncapobianco/pyATS_MCP/testbed.yaml"
200      }
201  }
202}
203```
204🧾 Explanation:
205command: Uses Docker to launch the containerized pyATS MCP server
206 
207args:
208 
209-i: Keeps STDIN open for communication
210 
211--rm: Automatically removes the container after execution
212 
213-e: Injects the environment variable PYATS_TESTBED_PATH
214 
215-v: Mounts your local testbed directory into the container
216 
217pyats-mcp-server: Name of the Docker image
218 
219env:
220 
221Sets the path to the testbed file inside the container (/app/testbed.yaml)
222 
223 
224✍️ Author
225 
226John Capobianco
227 
228Product Marketing Evangelist, Selector AI
229 
230Author, Automate Your Network
231 
232Let me know if you’d like to add:
233 
234A sample LangGraph graph config
235 
236Companion client script
237 
238CI/CD integration (e.g., GitHub Actions)
239 
240Happy to help!
241 
242# The testbed.yaml file works with the Cisco DevNet Cisco Modeling Labs (CML) Sandbox! 
243