Anthropic Claude Ecosystem Explained: Projects, MCP Servers, Skills, Sub-Agents

Hi everyone. 2025 was the biggest year of change in my life, both in my career and in my personal life. That kind of shift meant I genuinely didn’t have much time to upload for a while.
And looking at 2026, things still feel visibly turbulent for me, but I really don’t want to quit halfway. I want to keep this channel going and share small, practical takeaways from my daily life and work, mostly around AI.

It’s been a while since I made a video, so I’m a bit rusty. Once the momentum builds, I really believe I’ll get better and more consistent again. So thanks for sticking around

In this video, I break down the Anthropic Claude ecosystem using a simple office analogy. I explain what each piece is for, when to use it, and how they work together.

What we cover:

0:00 - intro
0:43 - Projects: persistent context for your docs, brand voice, and SOPs, plus the prompt caching benefits
1:47 - MCP and MCP Servers: how Claude connects to real tools and live data, like Google Drive and databases such as PostgreSQL
2:56- Skills: “progressive disclosure,” meaning you only load the right instructions when you need them
4:04- Sub-agents: specialized contractors with separate context and safer permissions for higher-risk tasks

I also walk through a couple scenarios to make it practical, like writing a brand consistent welcome email versus extracting invoice totals into a formatted spreadsheet.

If you’re building workflows with Claude and you’re not sure what to use, leave a comment with your use case. I’ll try to cover more concrete examples in future videos.

Links mentioned
https://claude.com/blog/skills-explained
(Skills explained: How Skills compares to prompts, Projects, MCP, and subagents)
https://github.com/VoltAgent/awesome-...