Get good in the terminal before you build an AI agent A newer member asked me today whether they should build an agent. They had just started using Claude Code a couple of days ago. My answer, in order: 1. Get really good inside the terminal first. That's where AI is most powerful. Give it access to your tools there. 2. Then n8n, which is what we use for automation. 3. Then an agent. A Hermes agent is very useful, but only once you know how to give it skills, tools and scheduled tasks effectively. They do different jobs, which is the part people mix up. An agent is always on. It runs in the background on a computer, you can text it when you need something, and it runs scheduled jobs. What agents aren't great at is trigger-based work. Email comes in, do this, respond, send a reply. Anything like that, we use n8n. The nice part now is that Claude Code is very good at building inside n8n. Give it the n8n skills plus a connector that lets it talk to n8n, and it can build workflows and debug them. You don't even need to log in to n8n. You describe the automation flow you need and it does it. Which one did you set up first?