Agent OS + Obsidian Skill + Local FREE AI!

**Julian Goldie** (0:00)
Today, we're going to be answering some of the latest questions about the Agent Operating System that we've built, how to use it, how to build them, how to get the most out of them, and what it can do plus the best ways to configure them. So we're going to get straight into this. This is based on the questions we get inside the AI Profit Boardroom community. We're going to get straight into this and start answering some of these questions. So one of the first questions we got was like, how to configure each of the CLIs inside the Agent OS when you set up. So basically, if you're getting our system, and you can set this up yourself if you want to. But if you get in our system, there's a folder with the CLI configurations, and you just ask whichever agent you're getting to set up to configure each of those CLIs. Now for most people, they won't use all those CLIs. So you can hide the ones you don't want inside the toggle in the top left, and then from there you can install the ones you want. Now, for example, if you're getting Claude set up for you, you can just say to Claude, hey, set up the Claude CLI inside the Agent OS, check that it works, and then if you have any others to set up, you can go from there as well. So if we have a look at the installation set up here, section number seven is the Agent CLIs inside the install folder. We've got another question here from Brandy, which is how can you set up an Agent OS and configure it when doing work for clients?
So I would configure that inside your memory system. So for example, if we have a look at the memory inside Obsidian, you would have different projects and different folders for each of the projects you're working on. So for example, when I'm doing work for the AI Profit Boardroom, we have a separate folder for that. When we're doing work for our funnels, we have separate folders for that too. And so for each product or for each client, you can have separate folders inside your Obsidian memory, and that way you can easily categorize between them. The other option is that you can build custom workflows for each client. So you could ask Claude, okay, I want to set up a SEO workflow for client A, and then you just configure it that way.
He's added a series of skills to make Claude Code autonomous, but it's made his setup super slow.
Yeah, I think there's a tendency sometimes for people to give too many rules and too many skills to their AI agents, and that makes them too slow. If you want your agent to be autonomous, like for example Claude Code, and you want it to work on a project where you just hand it off, you can actually use something called forward slash goal mode, and you can give it the goal, and then it runs autonomously on a loop to achieve that goal, and then a judge will actually check the work on every round and make sure it's actually done. So for example, if we go into Codex as the same sort of situation, you can go to goal mode inside Codex, you can give it a goal, and then it will loop round on every round until that goal is achieved. And the way that AI knows the goal is achieved is by checking the work with a separate model, a judge, and that makes sure the work is actually done. Otherwise, if you give your agents too many skills, too many configurations, too many rules, they can use up a lot of tokens and also be super slow and just not get anything done. The next question we got was, is it best to have local AI and Agent OS running on separate machines? So Scott was saying, you know, I have a computer just running local LM01 machine. And should I set up Agent OS to run on another separate machine that I do my work and need to keep running 24-7 on both?
So for me personally, I would have them both on one machine running locally. So you can have Agent OS running on your machine and then the local models running on the backend of the Agent OS. And that way it's easier to organize and also it's just running locally, so it's easier to configure as well.
Also for security, I think it's better to have everything set up locally rather than giving access to multiple machines with that sort of access. Blake was asking, you know, what's the best way to connect each CLI or just connect in general to Agent OS to bring on Claude, Hermes, notebook and so on? Are there any pre-built plug and play agents that we can put into the Agent OS for various jobs? And if so, where to access and upload them?