theme you see throughout the workflow. He's not trying to remove safety. He's trying to remove unnecessary friction. That's a recurring pattern in how he uses Claude code and how you should be as well. This is where Claude stops becoming just a coding tool. Boris explains that Claude can use MCP's model context protocol tools...matters because now Claude isn't operating in isolation. It has access to the same information humans use to debug, review, and make decisions. At this point, Claude isn't just writing code. It's participating in the engineering process. Boris says that this is the single most important thing for getting good results. Claude needs a way to verify
Поиск по транскрипциям
videos that you really don't need to code anymore. You don't need to know how to code anymore, because an agent will write it for you. Claude Code, Codex, Cursor, whatever it is, someone else will write the code, so you don't actually have to know things too deep. And I find this deeply, deeply troubling...worse. But once you take a step deeper into this conversation, this becomes extremely problematic. So, first, let's address the elephant in the room. Yes, LLMs, Claude Code, Codex, whatever, they keep competing with each other. And like now Cursor released their own model, I think, as well. And they become better, bigger, faster. Well, Opus and Claude tends
asking for legitimate assistance something else worth mentioning is that along with Claude 37 Sonet anthropic is also introducing a command line tool for agentic coding called Claude code it's available as a limited research preview and enables developers to delegate substantial engineering tasks to Claude directly from their terminal Claude code is an active collaborator that can search
think. First, go to app. supmemory. ai and create an account. Takes like 2 minutes. Then, you grab your API key. You add it to your shell profile so Claude Code can access it. Then, you install the Claude Code plugin from their GitHub repo. You just clone it, copy the skill folder, and restart Claude Code. That
agent that will review and optimize all my code to make sure it's efficient. And I'm going to hit enter on that. And now Claude Code's going to build out this sub agent for us. And this is a sub agent. It will be able to spin up whenever it wants, whenever it makes changes to make...show you here, we're just going to build the code review sub agent out. But this is also a great thing if you want to go into Claude Code and say, "Hey, I want to build some sub aents based on what you know about my app. What are some ideas you have for sub aents I can build
show you. But, let's get on the same page and talk about what they are. skills makes claw code insanely good at very specific repeatable tasks. So claude code is very much a generalist AI agent. Claude skills makes it a specialist. So if you have workflows or things you do that you do over and over again like...designing frontends or implementing Stripe, you can use Claude skills to make Claude code so much better at those repeatable workflows. And I'll show you a bunch of examples in a second. It saves you context which really sings you money at the end of the day because Claude skills are really memory efficient. Typically what people were doing
agent two over here, agent three over here, and then you'll kind of get their feed. I will note, I've done this before, unfortunately, because Claude Code renders your, it's not just like a simple text terminal. Basically they're like rendering this 2D image constantly on your screen. It can consume a fair amount of compute...websites back into just like we did with subagents, sort of that main thread. But what's cool is these three different agent teams, since they're all individual Claude Code instances, they get to do a variety of different things. They also get to access their own agents, use their own code base and stuff like that. So what
three ways to use Claude Code. A IDE extension. So an extension like in Visual Studio or cursor, the desktop anthropic claude app as well as the CLI which is kind of the classic way of using it which is just right in your terminal. There are pros and cons to each. I prefer the CLI and for this video...months and recently changed. The extension has a great guey, a great user interface which makes it very easy to use. It has all the basic features of Claude Code. It just lacks some advanced functionality we'll talk about. Then there's the desktop app. This is the Claude desktop app that is actually really cool and really neat
here's where it gets really interesting. People figured out you can run Claude code inside Anti-Gravity's terminal. Wait, what? Yeah, you can combine both tools in the same workflow. And this is why everyone's saying the combo is insane. So, here's the workflow that people are using. You use anti-gravity for the big picture...Anti-gravity's agents handle all the heavy lifting and automation. They plan out the whole project. They execute tasks. They run tests. All autonomous. Then you bring in Claude Code for the front-end precision work, the user interface, the design components, the debugging. Claude Code is just better at this stuff. It's more precise. It thinks through
just a note-taking tool that's based on Markdown. And then I have two terminal windows on the right. And the big thing that I use Claude Code for like I started using it coding just like everybody else most that's most people's entry into it. Um but I what I really loved about Claude Code with...coding is it's almost like you're pair programming with Claude. And I was like wow how do I pair with Claude on everything that I do? And so I started doing this for writing and for strategy and for literally all of the work that I do. This is my setup. This is my work obsidian vault
admin email to my. env file. When I type in /admin/ moderation after the local post, nothing happens. Thoughts on dangerously permission mode for claude code? Do you think it's too risky to use generally? No, I think you should. I think it's totally safe to use. I've never used I've never had Claude Code break...something where I'm like, "Holy crap, it just ruined everything. How could it possibly do that? " That's never happened. I'm totally fine with it. I trust Claude Code endlessly. I think one reason is because people expect you have the same needs as they have, so you must have the same workflow. Ignore the noise
HyperWhisper. There's a coupon code down below for that if you're interested as well. And basically I have Claude Code make a bunch of changes. And then I give all the changes including the summary to Codex CLI. And then I ask Codex CLI to come up with a critique of all the changes that were made...then I pass this critique back to Claude Code. And I say what do you think of this? And then I give all the critique over here. And the recommended changes and all. And then it comes up with a new plan after investigating everything that was done. And then I tell
first. It was like kind of the first development environment uh powered by AI and it is a really powerful tool and it's great for beginners. There's Claude Code. Claude Code got released earlier this year and it is point blank the best AI vibe coding tool out there ever made. It is amazing. It is mind-blowing
particular task. Next up is using ghosty as your terminal and then customizing it to use claude code better. So the cloud code team personally loves using ghosti and one of the ways that I have it set up. I can make a brand new tab pretty easily by doing ctrlt and then also rename any tab using keyboard shortcut...combine this with hooks as well. So often I'm working on many different projects at the same time. So I'd have like five tabs open and often claude code would be done on one particular tab or require permissions but I don't know which tab to actually go back to. So what I have
about to become five times more productive. Anthropic just released a new feature for Claude Code that I've been waiting for forever and I'm sure you have too. You can now claude code natively on your phone on the go wherever you are and have the projects built out on your computer. No need to do any sort...your phone. You can now run projects from your phone that build on your computer without having to do any sort of technical work. It's official. You can claude code anywhere. If you're a nerd like me, you can officially touch Grasp while also still cla coding. In this video, I'll cover how it works
This is going to change everything. So, warp code just dropped and this is the biggest thing to happen to code in since chat GPT. Everyone's been using Claude code for months and suddenly this new player shows up and makes it look like a toy. The coding world just got flipped upside down. All those expensive developers charging...more waiting months for developers. No more paying crazy fees for simple changes. No more being held hostage by technical debt. Now, the pricing is really interesting. And Claude Code charges per conversation. With big projects, those costs add up fast. People spend thousands in a single month. Warp Code uses a monthly fee and you can build as much
Another thing related to Claude Code is that DeepSeek drops another model, DeepSeek-V3. 1. And they say it's better at multi-step agent tasks, which means that it's going to be better at coding. And they actually made a Anthropic API endpoint, which means that you can now use DeepSeek in Claude Code and benefit from...features that Claude Code has available but using DeepSeek. And I go through how to do this in my previous video. It should be linked above somewhere right now and also down below in the description. Another relevant thing for Claude
exactly follow this. Copy literally copy. You don't have to understand what it says, but if you want to understand, it's just telling your computer to download Claude Code. Mine's already downloaded, so I'll show you the next step, which is in your Obsidian, go down to settings. Go to community plugin. If you haven...enabled it, turn it on. And then go for browse and check terminal. I found this to be the best way and really easy to understand way of using Claude Code inside your Obsidian and just download it. I already have it downloaded, so it's already enabled. But once you enable it, you'll be able to see this
development server. And typically what would happen is Claude will start the development server and then it'll get tied up. You can't really do anything with claude code while it's running the server through claude code. So I'm going to hit enter on this. But now what's going to happen is I'm going...about again. You can keep your workflow moving seamlessly. This is really, really helpful. Will save you a ton of time. So, we went through five new features that Claude Code just released that they don't really document anywhere. To be honest, I think the status line one's my favorite because that Pomodoro timer has kept
know push changes to GitHub with clear summaries and this is a video that you're definitely going to want to watch if you're someone that uses Claude for coding should we be doing like big smile or uh no big smile is creepy that's sort of way think I'm Boris I'm an engineer...product manager we love seeing what people build with Claud especially with coding and we want to make Claud better at coding for everyone we built some tools one of which we're sharing today we're launching CLA code as a research preview CLA code is a agentic coding tool that lets you work with CLA directly in your