Welcome to How to Master Claude Code from intro to advanced.

Today we're gonna go through five introductory things that can up your Claude Code game. Let's get started. Number one is claude.md. Claude.md is like an instruction file that you can give to Claude in order to accomplish things the way you like them. This could include things like what you want to accomplish, the tech stack you wanted to use, as well as any styling preferences that you have. Now it's important to know that there are actually three different levels of claude.md that you can have. At the very top you can have it at a user level, which means all your development and styling preferences can go in here. Second, it can be at a repo level for your entire project, and third it can be at a sub repo level. One thing I really like to do is before I start developing anything, I chat with Claude to flesh out all my ideas as well as the tech stack. And then at the end of this conversation, I tell it to write everything into a claude.md such that it can be fed right into Claude Code to start development. This is also very good for saving space in your context window or if you're closing out a session and resuming a new session but want Claude to remember what happened. The next one is model selection. If you're noticing that your tokens are running out really quickly, it could be because of the model you're using. So Claude has three different models ranking from the most powerful one is Opus which is also the most token consuming one. Second one is Sonnet which is kind of the everyday regular all purpose model and then there's the Haiku for very fast response and less intensive on token consumption. So ideally a lot of the thinking and the planning is done with Opus, a lot of the execution is done with Sonnet and some of the quicker operations or smaller questions can be done with Haiku. Next one is plan mode. This one's super critical if you're about to do a big change on your code and you want to think through these steps with Claude Code before start doing things. Essentially you're asking Claude to lay out a plan of how it plans to execute this change and because this is all laid out, you can provide feedback, you can recommend changes, you can ask it to do things a little bit differently before it actually executes on this, which means you don't have to redo things and waste a lot of tokens. Point here though, you don't actually need plan mode for small tasks or if your change is not complex because if it's very straightforward or if it's very directive, then Claude can actually just do it. The next one is how to interrupt your Claude Code halfway through running things. This is especially helpful if you've noticed after you submitted a command that something is incorrect. So rather than waiting for Claude to finish executing and then having to re-correct it, you can just hit the escape button on your keyboard to interrupt it and correct any mistakes from there. Next one is providing image references. I cannot stress enough the importance of providing image illustrations of the changes you want it to make. Circling, writing, drawing, highlighting things and then providing that image to Claude Code for a change is much easier than describing, oh that button on the top beside whatever whatever or that thing at the bottom, change it to this. And this could be as simple as just a screenshot that you can drag into Claude Code or you can mark it up however you want. So that's it for the intros, next video we'll go through some of the more intermediate actions that you can take in ClaudeCode.