Working with an LLM

If I need to lab something, most of the time I have an LLM generate the configs

Two reasons:

• Troubleshooting practice
• Learning the minimum to design a solution

Low tech method

• Drag and drop the nodes
• Wire the topology
• Show the physical connections
• Take a screenshot
• Point the LLM at an instruction link
• Ask for the config files
• Drop them in

High tech method

Describe the topology at a high level with design constraints.

• Solution being tested
• Protocols used
• Pass/Fail conditions

Tooling:

• PyATS
• CML MCP
• CML Breakout