How I Automate 80% of my Social Media Content with Claude Code

All of your social media content generated from one single prompt and ready to post in one click. What's up guys, in this video I'm going to be showing you how you can generate all of your social media content from a single prompt using Claude Code, ready to post with images, captions, hashtags, everything, whether that's Instagram, LinkedIn, Facebook, or any other social media platform. Just quickly, if you don't know who I am, my name is Rory Ridgers and I run an agency where we help business owners eliminate operational bottlenecks in just 14 days using AI. So if that's something you might be interested in, I'll leave a link in the description of this video and you can book up a call with me. But without further ado, let's jump into the video. Okay, so we're inside of Claude Code now, and there's a couple things I want to show you inside of this workspace before we jump into actually using this system. So first of all, we just have this Claude folder here, and in here are going to be our commands that allow us to actually do everything. So if we open up these commands, you see we have a couple in here, and what each of these actually do is pretty self-explanatory based on the names of them. So First of all, we have this content command here, which is going to do exactly what I just showed you. Create content among various different social media platforms, ready to post with images and text. And then we have some other commands here that just are going to do those individual functions on their own. So you can generate carousels from one command, Facebook posts, images, Instagram posts, and LinkedIn posts. And all of those are going to come with the images and the respective copy. And then repurpose as well. You can give it YouTube videos, you can give it blog posts, articles, news posts, anything. It's going to repurpose that into content for you. And then another super important thing I want to show you is all the context. This is one of the things that makes this system work way better than just prompting ChatGPT and asking it to generate you social media content. Actually has full context on me and everything that I do. So if I go brand here and I go to audience, you can see here this is going to be my audience. So who I'm trying to target with my social media content. You can see we have SMB founders and operators, what they care about, And then also secondary audiences like aspiring AI builders and other agency founders, and then what they care about. So basically just who I'm actually talking to in these posts, and this is specific to me. Then also some examples here. So it's taken some of my LinkedIn posts, my Instagram posts, and then also Facebook posts and just carousels. And it can use that to get a good tone of voice for me and the type of stuff I like to talk about in my posts. And then it just has some examples of hooks in here. So I've written out all of these different hooks. That I would actually use so it can use them as hooks, and it's in my voice. And then finally, we actually have voice. So this is just going to be exactly how I talk, so it looks like it's actually coming from me. Um, and you can iterate this over time. It's probably going to be a little bit of a longer process, but this file just basically allows it to make it look like these posts are actually made by me and not just generated using Claude. And then we have the outputs folder, which is just where all of those outputs are actually going to go and be stored. and then we have this review folder here. And basically what this does is as the system goes through, it's gonna be reviewing everything it does. So it has some static stuff in here to basically review all the styling, make sure it's on brand. And it also has some templates in here, and this is gonna be for that landing page that I just showed you. So it's gonna generate that full landing page as well. And this is basically just the template for that landing page. So it just looks the same every time. And then we just have some Python scripts in here. Well, we just have one Python script, which is to generate those images. That I showed you. And by the way, this is using the new GPT image model, GPT image model 2, I believe it's called. But it's the new one that came out a few days ago at the time of recording this video. So this is the Python script that's just going to call that function. And this is really the only, um, sort of external tool that it needs, this Python script, to generate those images. Of course, it's also doing things like looking things up on the web and scraping news articles and things like that, but Claude Code can just do that by itself. And then we just have a.env file here that just simply has my OpenAI API key in to be able to generate those images. That's the only API key we need for this. And then just to tie it all together, we also have the Claude.md file in here that just gives it a little bit of context on who it is and what it's actually doing. And then the one rule, always load basically these brand folders, so these context folders, so voice, audience, and hooks before it actually starts writing anything or planning out any content. And then we just have the workspace structure for it here. This is really great to do in the Claude.md files because it means that default, when you just send a message to it, the first thing it's going to read is the Claude.md, and it means it doesn't have to actually look like into all of these files into the workspace to see where everything is. It can just look at this and know, okay, I need to grab something from context. I know exactly where that already is, or it needs to generate some images. It already knows where to find that Python folder. And also just some guidelines on how to actually run the content pack. So basically just like a workflow in here of what to do when I tell it to do stuff. And then just a review section here that I was telling you about earlier, so it can self-review everything that it generates. And also generate a landing page to display everything. And then a quick voice reference just in case it needs it for little things. And then also just the environment that it has here. So we're just going to get into a demo here. We're basically going to get it to generate some social media content for us, and then I'll just show you what's happening as it's running through. So we're going to run the command content. As you can see, it pops up here, and we're going to say content. We're not just going to run this though, we're going to give it some more context on what we actually want to create the content about. So I said OpenAI dropped their new model 5.5 yesterday. I want to create some content on that. So OpenAI dropped their new GPT-5.5 model. When OpenAI drops anything like a new model like that, it's always going to be in the news for like a day or two. So we want to jump on that and make some content on it. So we're going to send this off here. And as you can see from its ClaudeMD file, the first thing it did was it read the voice file, read the audience file, read the hooks file, and then the examples file. So all of these context folders it read them and it reads them instantly. You don't need to think, 'cause there's like 4 in there, it's going to take a little while and kind of clog it up.' Now it's preloaded with all the context on exactly how I speak, who I'm targeting in my content, some examples of my past content. That way it's going to be able to plan out and write much, much better content for me. Then it's just come back to me and said it's loaded voice, audience, hooks, and examples. It's also grabbed today's date, which of course is pretty important when you're creating social media content. Now it's drafting a pack with the angle of GPT-5.5 drops. I'm not switching because the model was never the bottleneck. The system around it is interesting. Basically calling out the founders chasing specs instead of building workflows. And now you can see here, it's just actually gonna start creating all of the assets for that. So it's written the Instagram post here, written the carousel post, the LinkedIn post, the Facebook post, and then it's also written the image brief. Image brief basically just means the prompt for the, uh, images there. It's generated that hero image there. Now it's writing the summary of basically everything that it's done, and then it's going to load us up a full landing page where we can actually see all of this content. So the way image generation APIs a lot of the time work is you kind of send off a job and then you have to poll it to make sure it's actually done. And if it's not done, then you just poll it again, like, you know, 30 seconds later or something like that. You can see that's what it did here. It did a bash, which was check if image has been generated yet, and then it said verify hero image was created and VueGen output. So we just same thing again. And then that time the hero image had been generated, so basically it moved on with the next stuff, which was launching that landing page for us to actually view. So you can see it's now done everything. It's given us all of these assets here. And if we was to go into outputs, we can actually see all of these outputs. So the summary here of basically everything that it generated, the angle, the notes, and then the LinkedIn post is here, the Instagram post, the image brief. So basically the prompt for the image, the image here. Cool. And then the Facebook post and the carousel. So let's open up this landing page so we can actually see it. I mean, the way that I showed you at the beginning of the video. So as you can see here, this is the Instagram post that actually visualizes it for us, how it would look on Instagram, all of that, the likes, the comments, the save button, everything like that. So we can see exactly how it would look. And if we wanted to actually use this, all we do is we just hit copy caption, and then we can just download that image as well. Then that downloads there. Then we also have the carousel, which is GPT 5.5 dropped yesterday. I'm I'm not switching. And then we can flick through each slide in this carousel. So we can choose to post something like this, probably on LinkedIn or Instagram. It would work best. You can see here, nice, like clean, concise slides there on the carousel. And then we have a LinkedIn post. Again, visualizes it exactly how it would actually look on LinkedIn. So we can see. And then we also have the Facebook post. So I think with this, I like it. I think the copy looks good and I like the angle, but I think this image could use some work. The image looks great. It looks really realistic and stuff. But I think it should actually contain the ChatGPT logo or something to do with 5.5. So we're going to go back over to Claude Code, which is going to tell it that. So I just said I like the image, however I wanted to actually contain something relating to GPT-5.5, like a logo or a graphic. So we're going to send that off and it should just be able to change that image for us, do a little bit of research, come up with an idea, and then just send off another request to generate a brand new image and then just basically replace it for us. So it just said good call, the blurred tech news framing reads as generic. I'll reprompt so GPT-5.5 is clearly visible on screen/graphic, but keep the editorial calm composition so it still lands the not switching thesis visually. Updating the brief and regenerating. So actually, we can look in here. It has updated the brief. It's updated it. So an editorial overhead photograph of a minimalist wooden desk in a soft morning lighting is pretty much what we had before. But then it's saying behind it, a clean white laptop screen clearly displaying a bold minimalist launch headline in large black sans-serif type. GPT 5.5 5. The headline text is Chris, legible, unmistakably the graphic focal point in the upper third of the frame. Awesome. So we just sent off that request. It said generation kicked off. I'll wait for it to finish. So it's just going to do that polling thing again, which you're going to check back in, see if that's image generated. If it hasn't, it will pull back in like another 30 seconds. Okay, so I got a little bit impatient and I just said, is it done? And it checked the status and said, no, not yet, still generating. And basically just said, chill out, I'll let you know when it's done. So I did that, just told me to hang tight. And then it got the notification saying it was done. So it grabbed that image. It says we can just refresh the page and there'll be the new image there. So if we come over, hit refresh. Yeah, as you can see, GPT-5.5. So awesome. It's got the image there now. We've actually got GPT-5.5 in the image and it's using the new GPT image model, which is really nice. So obviously this is an example of just generating one post when a new piece of news in the AI space lands. But of course you could get it to literally generate you a week's worth of content or like 2 weeks or a month's worth of content if you want it to. It really doesn't make a difference to the system. The only thing you're paying for is that OpenAI image generation model, which, I mean, for like 10 posts, it's really not going to be expensive at all. And it also generates this nice landing page where you can actually view everything. In the other systems I've done where I do a similar kind of thing, puts all of it into an Excel spreadsheet with like the title in one column, the caption in another, the hashtags in another, and it's all right, but you can't actually see how that's going to look. With this, you can literally see every single time exactly how this post is actually going to look across different social media platforms, which is absolutely awesome. Okay, that is it for the video, guys. I hope you enjoyed it. Like I said at the start of the video, if you were looking to see how you can implement AI into your business, and I've left a link in the description of this video where you can book up a call with me. So thanks for watching. I'll see you in the next video.