claude-mcp-pipeline

In February 2025, I developed a new AI Agent, Claude + MCP, helping me data analysis and file organization

Here are the questions I'm interested in answering:

1. Why do we need to ask, copy, and paste code into other tools just to test if the code works?
2. What does next-generation analysis look like?
3. How can we develop our own AI agent?

I took the following steps to create my AI agent:

1. Downloaded Claude MCP and SQLite
2. Modified claude_desktop_config.json
3. Set up my agent's to become data analyst and file manager in the JSON file

Here is my demo:

1. Data Analysis – No need to use SQL or Excel to generate plots

https://www.loom.com/share/d3e1c0f0203642dea0ff3d63f0e9b96e?sid=dad16c08-cbc8-4571-8cd9-952304fd3ac9

2. Filesystem – Files are moved and renamed through AI communication

https://www.loom.com/share/d694318e1b104aaca72ed02012e0c857?sid=ed1681ba-6320-4b9b-a66c-0befcaf44615

Claude MCP helps streamline the workflow and saves us time, allowing us to focus on the structure of analysis.