Transcription
MCP or model context protocol is the essential kind of interface and building blocks for extending AI tools that you already use and know and love to make them exponentially more effective at accomplishing the tasks that you want them to do. And in this video, I'm going to walk through the eight MCP servers and connections that I use all the time from development to organization to security to marketing and ops.
Hey, I'm Craig Huitt. Welcome back to 100 days of AI. Let's dive into the first MCP that I use all the time and it is Playright. So here we see in the GitHub repo from Microsoft. So Playright is an MCP kind of sponsored and maintained by Microsoft and the purpose of Playright is really to give things like cloud code eyes into the browser so it can see what's going on on the front end because I think one of the criticisms of AI systems and coding agents like cloud code is like, "Hey, it writes really good kind of backend functional code but it can't see what it's doing on the front end." You can't say like, "Hey, make this look like, you know, the Basecamp website or make this look like the Stripe website or make this look like Google or whatever." It just can't see that. Playright gives Claude code eyes so it can see and iteratively loop through things until it achieves what it wants. So here on the GitHub repo has really straightforward instructions for how to install this. I'm going to do a deep dive on Playright as an MCP and Claude Code soon. So subscribe if you haven't already to get the deep dive on that. I think you're really going to love it.
Okay, next one is Ref. And Ref is all about getting context of documentation of tools that you use in development. So this is a software play for sure and you might have heard of Context 7. Kind of similar uh tool as Ref, but Context 7 pulls in all of the documentation anytime it's making a call. Ref only pulls the data that it needs for the service, you know, like these mentioned here and a whole bunch of others when it needs it. Right? So it says it decreases the token limit and the token usage by like 80 or 85% over something like Context 7. So if you're someone who's doing development using a bunch of external services like these or the thousands more, then you're going to want something that pulls in docs and information on how to do effective uh coding well for these external services you're integrating with. Ref is going to be the MCP that you want to use for that purpose.
Okay, the next one is sequential thinking. This is also in a GitHub repo here. And sequential thinking uh allows you to think in a more kind of structured and sequential, you know, by the name of it, way. And so what happens here is I think a lot of times we give tools like cloud code really complex tasks and it can't think linearly through the progressions even if it's making like a checklist or something uh to achieve the task that we give it. It kind of jumps around, makes some assumptions, then arrives at the end and you and it don't know how how it did that, right? And sequential thinking is just a tool to give more structure, I think, to the thoughts and the process of decision-making and putting together pieces of information for you uh in something like cloud code or an agent than just without it, right? And so you can see here the tool and the inputs that you can give it. And again, like it has instructions here for installing cloud desktop using VS Code and then building this and using it. So sequential code or sequential thinking, I think definitely a tool for any application, not just coding, but like marketing, task operations, uh sales, all these kinds of things, right? So sequential thinking, probably a must for basically anybody.
Okay, the next one is back to the development world and is Sim Grep. Uh I know that Claude has built-in security uh kind of agents that can and do run inside Claude Code, but I prefer a little more specialized tool in Sim Grep because I just think that anytime somebody focuses 100% of their attention on a very specific task, it usually ends up better than even an amazing tool like Claude and Claude Code uh can do, right? So you want to, if you're taking security seriously, which we all should, especially in vibe coding, uh because I think that something happens in vibe coding is like, "I'm not a developer." I don't, and even if you are a developer, I don't think you're probably paying attention to all of the code that is being written. And so you need to rely on an external service like Sim Grep or even the built-in kind of Claude Code security uh review agent to do thorough reviews of your code. I would say even if you're doing manual coding, like old school manual coding from a few years ago, something like this is a great service to go in there and just analyze your codebase so that you don't have to rely on external security researchers or, God forbid, some kind of vulnerability that's exploited uh and you expose a bunch of customer data or your app goes down or something. So for security, the way to go.
Okay, the next one is Stripe. And Stripe, obviously, payment processor to kind of all of the internet. I think just an amazing company. We've used them at my company, Casto, since day one. And what's cool about the Stripe MCP is it lets you interact with your account to do a bunch of stuff that you might have to do manually, right? So, uh, it has here like the tools that it can use. So, account, balance, coupon, customer, dispute, invoice, payment, link, price, product, all these kinds of things, right? So, if you're building an agent or a tool that needs to interact with Stripe on a regular basis to do stuff that you might go into the Stripe dashboard to do manually, this is a cool way to not have to like hardcode things in, but you have an agent that can use this MCP to call Stripe uh and do things in a dynamic way that aren't kind of hardcoded in. So, if this kind of fits your use case, I found this really helpful. Even just having background information for like a product that I've integrated Stripe into, it just gives me a little clearer picture of what's going on in the product and in the integration with my payment processor. It's worth checking out.
Oh yeah, the next one is Supabase. Uh, and I like this because if you've vibe coded at all in something like Bolt or Lovable or Vzero or in Cloud Code, you know, the default that it goes to is usually a Next.js app and Supabase. Uh, so Supabase, just for context, is like a cloud-based database provider that integrates really well with Next.js as a framework and the Supabase MCP server is just a cool way for you to abstract away what's going on in the backend, how your data is organized, how tables are set up, all these kinds of things. It's really a nice kind of uh abstraction and way for you to interface and visualize with what's going on in the backend. Again, especially if you're not a developer and you're like, "Hey, I don't really understand all this backend stuff. I need a way to more clearly kind of see and have my uh AI coding tool like Cloud Code talk to the database and understand what's going on there." It's going to set up things like databases and tables and indexes and really optimize performance of your backend and your database uh especially for a Next.js application. Cool.
So the next one is definitely a marketing one and it is Appify. So, if you're a tool that needs real-time information, so you're scraping LinkedIn or you're scraping Reddit or Instagram or you're scraping particular websites, uh, Appify is the way to go. It has a ton of built-in. I'll just open, um, I'll just open Appify. So, you know, the headline here is "Get real-time data for your AI." Appify has all of these kind of recipes or they call them actors for ways to very easily scrape data. It works on kind of a token and usage system, but if you're like, "Hey, I need to go get real-time information from Twitter or Facebook or TikTok or whatever." Uh, you can do this really simply with Appify. It is not free. Uh, it does cost, but if you're like, you know, running a whole business on this, you can imagine like even paying $200 a month being really worth it to get real-time information into your application. Say like running a coupon, you know, program or something like that. Like maybe you need to go get a bunch of information from Amazon. You can do this with Appify and Appify's MCP server again just gives you a lot more clarity and visibility around like searching actors, details, calling an actor, documentation, even costs on these runs and how much all this is costing you. So definitely worth checking out if you're pulling real-time information from the web into your application for you and your users to see. Definitely worth checking it out.
Okay, the last one is a general purpose one but probably leans more towards ops and sales and marketing and that is Notion. So, uh, our entire company kind of runs in Notion. Everything from like an operations, sales, marketing, customer support perspective is planned out, executed, and tracked in Notion. And if your company is the same way, then this is definitely worth checking out. If you're wanting to like integrate an internal tool into the rest of your company, Notion is the very obvious way to do this. Is you build a little internal tool and you say, "Hey, like I want this to be able to talk to, you know, this customer service reporting tool that I built." Uh, like this is what you're going to need to do. So, uh, Notion, obviously, really robust, really flexible tool and the MCP server here uh just a really great way for especially probably like an internal tool that you vibe code up to talk to the place where the data and the information in your company lives. Uh, that's kind of how I use and how I think about the Notion MCP server is like, "Hey, uh, I have this tool that I want to like pull some other kind of external data from and I want the Notion data to be part of that. I would use the Notion MCP server to pull that off."
So these are the eight MCP servers that I use very often and I would suggest you checking out if you're getting into extending uh AI platforms and services like Claude or ChatGPT or Claude Code or an AI agent that you've built up. These are the eight that kind of span development, operations, marketing, product, and real-time data. Definitely worth checking out. Uh, if you have any questions, drop a comment in below. Please like, subscribe, smash the bell, and if you're new to MCPs and are wondering what the heck we're talking about and why MCPs are so important, check out this video right over here.