install Claude Code on a remote server so you can have it run dangerously skip permissions over there and do like web research for you and stuff like that. Without worrying about it accidentally deleting your files or doing unintended things on your computer. So I'm gonna log into my Hetzner account and then just quickly...choose the cheapest one, which is $3. 49 a month. Choose— choose Germany over here, create my SSH key and add to that here. And then call the server Claude Code and press create and buy. And now what I can do is SSH into that server copy over the install command for Claude code, go back to the server
Поиск по транскрипциям
problem is coding agents like Claude Code are not very visual tools. When you want it to explain something to you visually or even just help you explain something to someone else, they really don't do a good job unless you do a lot of prompting and give it the right abilities. And that is what I have been...working on. And so I've taken my entire workflow and I've packaged it up into a skill so I can use Claude code or any coding agent to help me build all these beautiful and practical diagrams. And so giving my coding agent the ability to argue visually. This is the primary instruction set that guides Claude code
technology behind Claude Code. This allows you to build agents just like Claude Code. So if you want to build your own agent that has access to tools so it can search the web, it can build files and create spreadsheets, it can write code, you can now build that for yourself. So if you have a very specific...your own custom agent using the Claude agent SDK, all you need to do is take the link for this site for the blog post. You can go into claude code and say, "Hey, I want to build a financial analyst agent using the Claude agent SDK. Here is the link. Paste that in. Hit enter. " And now Claude Code
allowUnsandboxedCommands`. And basically a problem that was happening with some people, for example, this person on Reddit, they were running sandbox mode, and then all of a sudden, Claude Code decided to manually override sandbox mode by running a bash command outside of sandbox mode. And how this works is that whenever Claude Code tries to use a bash tool...check against the sandbox to see if that command is blocked or not. Now, in some cases that can be useful, but also you don't want to risk Claude Code overriding the sandbox. So what you can do to prevent that from happening in your case is copy this over, then go to your sandbox. And then in brackets
chats, go to session memory, and then you can see a summary of that session. And basically every time you send a message to Claude Code or every time it responds to you, it always adds onto the session memory, the summary. So you can see it kind of looks like this. You have the session title, and this...title that Claude Code came up with itself, the current status of the session, which includes anything completed, any discussion points, and open questions. And then like a bunch of other things that you can see described over here, some learnings, some key results, and then a work log, which is automatically updated every time you send a message
honest for me, it's not super important. Just because I want it to be a little bit special and then show you the parallel capacity of Claude Code. What I've done here is I've actually opened up another anti-gravity instance. And what I'm going to show you guys how to do is actually design multiple...site to my specs. You don't have to develop in multiple tabs. This is something that I think you learn how to do the more of these Claude code agents, frankly, that you're orchestrating. The benefit to this is obviously you can develop basically however many times faster as tabs that you have open. But the downside
devs or anything like that. But I personally think my workflow just going to one of these websites and then copying the screenshot over and then moving everything into Claude Code is like way higher quality. But there are other cool ones here. Context 7 is pretty nice. Context 7 basically just allows you to search through...cloud. And then it can read it in a very token-efficient manner and do cool things with. None of these things I want to say are required. Vanilla Claude Code does really, really well without any sort of extensions or plugins at the moment. But you know, just worth us chatting briefly about that. And then there
many people out there that overcomplicate the hell out of this. So I'm going to do my best not to. If I open up a new instance of Claude Code over here, and then I type this backslash and then scroll down, you'll see that I have access to a bunch of really cool functions here...could find out for yourself. And so what I reckon you guys do right now, if you've never done this before, is head over to your own Claude Code instance without even watching any of this and just type backslash context and look at all of the things that are currently consuming your prompt. Now I should note that
project in claude code called sorry claude. This was my essentially marketing assistant today for going on your podcast. What I wanted to do was, so you can see here in my little notebook, I just collected all of the physical notes that I had on this promoter concept that I wanted to talk about with you at the start...simple version of it and I was like this is pretty fine. Then I asked it here. I said, "Cool. I'm going to move this project over to claude code. " So, I'm going to take the HTML file, but what I need from you is to create a cla MD file that gives you the claude code
might start to think about an outline and be like okay well the introduction needs to be strong hook around value of doing project work with claude code. Um then introduce like claude modes plan auto accept um reference safety article um walkth through detailed example interview coach. So let's see what it came back with. So I found...this in auto accept mode? Um, and so that's really nice, too. And I know this is a little meta. I'm writing a blog post about Claude code, but Claude can actually answer questions about anything, right? So, I could be writing a blog post about how product teams do customer interview analysis and I can still
skill. The reason why I love this skill is it helps you validate ideas before you build them out. If you're anything like me, you love building in Claude code. You love coming up with side projects to build out. But one thing many people don't do is they don't validate their idea before they build...create skills and it's going to build it out for us. It's now creating a skill. mmarkdown file. This is what we can plug into Claude and Claude code to teach Claude how to do things for us, including validating our ideas. Perfect. The skill has been created and packaged successfully. So, in a second, it's going
those tools, basically all of the time. And there's nothing you can do to fix that unless you wanna go in and make your own version of Claude Code or something. I will say, I think that some of these things are unnecessary. I mean, I definitely don't need the Jupyter Notebook calls. I think there...additional features here that maybe I don't need or we could probably make them smaller. But this is something that the Claude Code team is constantly improving, constantly pruning and so on and so forth. Next up, we have the MCP tools. Now, unlike system tools, MCP tools are things that you define yourself, which means every
there, but let's keep adding new things here. Let's think about how we can add more here. One of my favorite things to do in this new Claude Code for desktop app is take advantage of the fact that it has the chatbot built in. So, if I open this sidebar and then we click on this chat...your daily streak. What are other cool high retention and engagement features we can add? This is my favorite thing to do when using Cloud Code, especially Claude Code for desktop because it's so easily accessible. This chat feature is we're basically making this our co-pilot CEO and we're saying, "Hey, what other things
works 24/7. I can build entire applications now without hiring a dev team. Need a custom tool for your business? Claude builds it. Want to automate a workflow? Claude codes it. Have a client who needs a web app? Claude creates it. This is how you scale without massive overhead. But here's where it gets even crazier...insane. Think about your business right now. How many hours do you waste on manual tasks? How many processes could be automated but aren't because coding is expensive. Claude Sonet 4. 5 eliminates that problem. But here's what really blows my mind. This is just the beginning. Anthropic released this as Sonnet, not Opus. Sonnet is supposed
days ago from a company called Moonshot AI and the numbers are insane. On the hardest coding tests in the world, SWEBench verified. This thing scored 65. 8%. That's higher than GPT4. 1 at 54. 6%. It's almost as good as Claude Sona 4. But here's the crazy part. It's completely free and it's open...number one when using Claude 3. 5. Most coding agents struggle to hit 50%. This thing is legitimately crushing it. But here's what makes this different from Claude Code. Claude Code costs money. You need a subscription. Trade Agent
tasks here as well. Right. So this is showing me all the tasks from across all the different platforms I'm using Codeex on. Again, an advantage compared to Claude Code because Claude Code is just the CLI, right? So here we go. Add prompt optimizer tool to the toolbar. Right now it's not there. Right...commit this to GitHub, which is really simple to do. I basically have AI employees working for me whenever I want. Right? If I'm using Claude Code, if I think of something to do, I have to go to my computer, open up my CLI, and give it commands. With codeex, what makes this workflow really, really interesting
prompts at a regular interval and basically replicate all of OpenClaw inside of cloud code. Now, I did make a video about a week ago where Claude Code added a remote control feature whereby you could control Claude code from your phone, but that was kind of limiting because it was only with inside of the Claude...sure to make this approach secure. You add your user ID by getting cloud coded it for you. And now that is done. The benefit is I can get Claude code to do pretty much anything else on my computer via Telegram. So you can see it says done. Both scripts have been secured blah blah. So now that everything
Because they basically said that Claude Code excels in detailed conversational coding, but it hits its context limits with large projects. Gemini CLI, with its massive context window and planning capabilities, can digest entire codebases and provide strategic insights. And they basically gave the example that they were using a Flutter project and users reported that switching between tabs would...this actually felt like a system-wide issue that would sometimes happen on a profile tab, sometimes on Settings, but always after rapid tab switching. So they basically got Claude Code to then call Gemini CLI. And then Gemini CLI identified that some controllers were initialized in different ways on different routes over here. And then Claude Code worked
right away. Now, think about this. Before, if you wanted to build a landing page, you either paid a developer hundreds of dollars or spent hours learning code yourself. With Claude Incline, you get working code in seconds. And it's not messy code. It's clean. It's organized. It follows best practices. You can take that code, paste...into a file, and you've got a functional landing page. No joke. That's how good Claude is at coding tasks. This
this agentic interface that doesn't really take a lot of technological knowledge to run. Now, what I want to do next is run a simple request for both Claude Code and Gemini and compare the results. And then I think in the future I'm going to dig a bit deeper and follow up with a separate video comparing...Google account and you get to use the free credits right away. So what we're going to do is a quick little oneshot comparison between Gemini here and Claude Code here. And we'll be using this relatively simple prompt that includes a few components that would be interesting to compare. A single HTML file pomodoro timer with 25minut