All right, so this week Anthropic dropped a ton of new features across their products. There were a number of features that were released in Cloud Code. I debated going through and making some individual videos about some of these topics. Instead, what I'm going to do is I'm going to do a consolidated video just touching on a lot of what they recently just shipped. First up is remote control. Now, remote control, effectively what it allows you to do is now you can connect to your computer remotely from your phone. All that you need to do is with the latest version of cloud code is you can run / remote control and that will give you the ability to access that session of cloud code that's running on your computer on your phone or from a web app. When openclaw originally came out this, myself included was one of the key features that I found pretty novel within openclaw just being able to interact with my local computer from my phone. Next up was scheduled tasks within co-work. Now this is another openclawesque type of feature where if you take and stack rank some of the top features within openclaw you have the similar capabilities like remote control but also another very powerful feature that was built into openclaw was the ability to manage scheduled tasks. This can be things like having a summary or doing research at a certain interval, maybe organizing different files or following up on emails. There's a ton of different ways in terms of how people were leveraging this and this type of feature we've seen before. It was within chatgpt within their task feature, but now you're going to be able to run this directly within co-work. Now, another thing that they released were two new repos, knowledge work plugins as well as financial services plugins. Now, within both of these, so you're going to be able to leverage these both within cloud code or within co-work. And within each of them, they're structured a little bit differently, but effectively what they allow you to do is to have a number of different skills, whether it's within each industry, like they have within knowledge work plugins, or specifically for certain tasks within financial services. But to get started with these, you can go ahead and add the marketplace for the plugins, install the plugins that you're interested, and the nice thing with plugins is since they are built within natural language is if you do want to edit anything, it makes it very, very easy to do so. For instance, within the equity research, we have idea generation. If we take a look at the skill, you can see within here there is a particular flow that they've set up within theirs. But if you want to take these and fork these and make them your own, you can go ahead and do that. It makes it very easy just to edit the natural text within here. And the thing that's interesting with this is if you dig into each of these folders and these tasks, these would have otherwise been things that people would have otherwise done. Whereas now potentially for some or a large part of different tasks, you can have Claude accomplish a lot of these same tasks. So next up, they rolled out the new automemory feature within Claude Code. And now Claude can remember across different sessions different aspects about your project. So think things like whether it's debugging patterns, certain pieces of context, preferred approaches, and it will recall that and pull it into context very similar to something like skills. Now the difference with this, it's not built into the claw.md or skills. It's something separate in of itself. You can turn it on and off. And the nice thing with this and the way that you can think about this is this is effectively Claude remembering what it needs to remember. You can also prompt it and say, "Hey, remember this." It will go and write it to memory. But this is effectively going to be a file that Claude leverages itself. And what this allows you to do is instead of having these different approaches for memory, like previously I looked into continual learning where I had different skills that I would inject and have Claude remember certain things over time. And the nice thing with this approach is Claude will just begin to remember helpful things over time and you can always go in and edit it since it is a markdown file and you will be able to still leverage the claw.md as well as skills like you otherwise would. Next up, so they made some updates to the ask user question tool where now it can show you markdown snippets to display diagrams as well as code snippets. Say for instance, if you're asking to change a layout or create a particular front-end component, it can actually give you a nice little asky representation within markdown to give you different ideas on how it could potentially look. That's a nice two-dimensional way for a lot of different things, whether you're building a presentation or a website or whatever it might be. It can give you certain suggestions when it's clarifying things about the particular structure. And this is something that they're leveraging in their new web search capability as well as their web fetch. And you see it in a lot of different use cases. Think of how similar to cloud code where it will GP, it will search around for different code. It will write little helper scripts. It has its own shell environment where it can execute all of these different commands. It touches on a lot of those types of concepts. Now, additionally, one thing that I do want to touch on, now this did come out a little bit earlier, but 4% of GitHub public commits are now being authored by Claude Code. At this trajectory, we believe that Cloud Code will be at 20% of daily commits by the end of 2026. While you blinked, AI consumed all of software development. Programming has arguably changed the most over the past several months than it has in decades. And last up, Boris Jurnney, the creator of Cloud Code. He mentioned in the next version of Cloud Code, they're introducing two new skills, the simplify and batch skills. With the simplify command, it will use parallel agents to improve the code quality, tune efficiency, and ensure that the cla.md is compliant. And in terms of how you can leverage this, you can say, "Hey Claude, make this code change and then run simplify." And then for batch, you can interactively plan code migrations, execute in parallel using dozens of agents, and each agent runs in full isolation using git work trees, testing its workflow, and putting it up in APR. So, what's interesting with both of these new skills is it leverages a ton of the new capabilities. So, work trees as well as the ability to orchestrate teams of cloud code sessions in parallel. Otherwise, I'm going to put the links to everything that I showed you within the description of the video. But otherwise, if you found this video useful, please like, comment, share, and subscribe. Otherwise, until the next

Claude Code: NEW Remote Control, Auto Memory, Plugins & More

Show Full Transcript

Repurpose this content with AI