MCP Servers Directory

And to think the app was supposed to launch today. Instead, here we are — not only have I not launched it, but I’m sitting with a pile of doubts. I won’t lie — that’s what these posts are for. Maybe one day I’ll look back and read them with a smile, but right now, I have to admit it’s been rough.

On one hand, development is taking a bit longer than I hoped. By now, I should already have my first beta users. And the worst part? It’s not even about the development itself — it’s everything happening around the project and in my life that’s keeping me from dedicating the time I wish I could.

Sure, this is a side project, but my goal is to finish it, launch it, and with a bit of luck, help people build cool things with it. That hasn’t changed.

But here we are. The news that’s been on my mind lately: Zapier (yes, that Zapier — the automation giant) launched something “similar.” It’s not exactly the same, but it definitely stirred up a bunch of internal doubts. Who’s going to pick my tool when there are solutions from much bigger, more established companies? Now that these giants are entering this space, should I shift the focus of my project to users who aren’t covered by Zapier’s offering? Is Zapier’s launch just the beginning of a wave that will bury smaller players like me in invisibility?

And on top of all that… well, it’s what every founder hears sooner or later: 99% of projects fail. So is it worth pouring time into something that statistically has a 1% chance of making it? And now with Zapier’s launch (even if it’s not identical), that 1% feels more like 0.01%.

Maybe this project will end as nothing more than a few posts about something that could have been. Or maybe one day it’ll turn into a beautiful story. Maybe I’ll wake up tomorrow with fresh optimism and this will just feel like a moody Sunday post. But for now, I felt like it was important to share the real story — because building something isn’t as simple and glamorous as some people like to pretend.

This — the fear, the questions, the doubt, the uncertainty — this is also what building looks like.

Last weekend, I was supposed to launch the first stage of the project. I had it all planned, scheduled, and even told people about it. But… well, life happens.

That being said, I think I’m close now. If no more unexpected doctor visits get in the way (because yes, I ended up at the doctor, and yes, I should probably start checking my bike brakes before heading into the mountains), then tomorrow I’ll finally get Stage 1 online.

The funny thing is, I’m excited about Stage 1, not because it’s particularly amazing, but because once it’s up, I can move on to Stage 2, which is the one I really care about. It’s absurd. I’m racing through this first part just so I can get to the second part. Like when you eat your veggies quickly so you can get to dessert.

Anyway, I just wanted to write this down as a reminder that nothing ever happens the way we plan. If even the launch of a side project doesn’t go as expected, imagine the really big things in life. It’s almost comical.

But okay, enough philosophy for today. I’ll keep you all posted.

Edit: Also... finding the right domain name is becoming harder than developing the thing...

I’ve always had this dumb little fantasy of having a virtual assistant that does more than just answer me in a polite, British butler voice. More than once, I’ve found myself chatting with ChatGPT, Claude, or whichever AI was trending that day, thinking, “Man, wouldn’t it be nice if this thing could actually read through my documents, send a WhatsApp message, or do this or that without me lifting a finger?”

Well, turns out MCP servers do exactly that. (In case you had no clue what the hell those are, there you go, now you know.) And honestly? This is just the beginning.

I can see a future where AI integrates seamlessly into the stuff we use every day: our calendars, our emails, the software we work with, our browsers, everything. It blends in, sneaks into our routines without us even noticing. And that’s when we’ll finally have a real agent, something that actually saves us time on all the pointless little tasks so we can focus on more interesting things.

So yeah, this post is short, but I’m dropping it here as the starting point for this side project. The plan is to launch the first version this weekend, so if you are into AI stay tuned!

So, if you are readying this, thank you. Also, I see you might be interested in MCP servers but maybe you still don't fully understand what it can do for you or how do they work. The idea with this posts is to try to answer the most common questions and build from there. Thank again for being there!

1. What is an MCP Server?

Imagine your AI assistant is like a smartphone. Just as your phone uses apps to perform various tasks, an MCP (Model Context Protocol) Server acts like an “app store” for AI, providing a standardized way for the AI to access different tools and services. 

2. How do MCP Servers work?

Think of MCP Servers as translators between your AI assistant and other applications or data sources. They provide a menu of tools, each described in natural language with a defined structure, allowing the AI to understand and use these tools naturally. 

3. Why are MCP Servers important?

Before MCP Servers, integrating an AI assistant with various tools was like trying to fit a square peg into a round hole, each integration required custom code and was often cumbersome. MCP Servers standardize this process, making it easier for AI to interact with multiple tools and services without complex integrations. 

4. Can you give a real-world example of MCP in action?

Imagine you’re planning a trip. Your AI assistant could use MCP to:

• Check your calendar for availability.

• Book flights and accommodations.

• Send confirmation emails.

All these tasks can be managed seamlessly through MCP Servers without the need for separate, manual integrations. 

5. How does MCP improve security?

MCP incorporates features designed to protect sensitive data, helping prevent unauthorized access to external resources. 

6. What are some examples of tools accessible via MCP Servers?

MCP Servers can connect your AI to various tools, such as:

• File Systems: Securely accessing and managing files on your computer.

• GitHub: Managing code repositories and performing file operations.

• Google Drive: Accessing and searching files stored in your Google Drive.

• Slack: Managing channels and sending messages.

These integrations enable your AI to perform a wide range of tasks more efficiently. 

7. How does MCP differ from traditional APIs?

Traditional APIs are like individual doors, each with its own key and rules. MCP acts as a universal key, allowing AI to access multiple tools and services through a single, standardized protocol. 

8. Who is behind MCP?

MCP started as a project by Anthropic to make it easier for AI models to interact with tools and data sources. It has since become an open standard, with more companies and developers adopting it. 

9. How can I use MCP Servers?

To use MCP Servers, you would typically:

• Set up an MCP-compatible AI application (like Claude Desktop).

• Connect it to MCP Servers that provide access to the tools and data you need.

This setup allows your AI to perform tasks like reading files, managing emails, or interacting with other applications seamlessly.

10. Are there any security concerns with MCP?

MCP includes features designed to protect sensitive data and prevent unauthorized access, ensuring secure interactions between AI and external resources. 

I hope this helps, also, remember I am planning on launching my side project in the next couple of days, hopefully that will also help you on your journey navigating the MCP world.