Stop re-introducing yourself to every AI tool you use.
Every time you open a new AI chat, you start from zero. The AI doesn't know your name. Doesn't know your job. Doesn't know that you hate the word "synergy" or that you run a team of 12.
You spend the first five minutes of every conversation catching the AI up on things it should already know.
There's a fix. It's called an MCP server. And you can build one in about 30 minutes.
What This Actually Is
MCP stands for Model Context Protocol. Sounds technical. It's not.
Think of it like a briefing folder. You write down everything an AI assistant would need to know about you, put it in a folder, and then connect that folder to your AI tools. Now every AI you use can read your briefing before it talks to you.
No more repeating yourself. No more "I'm a marketing director at a mid-size SaaS company and I prefer a casual tone and..." every single time.
You write it once. Every AI reads it automatically.
Step 1: Write Your Briefing
This is the most important step. It's also the least technical.
Open a plain text editor (Notes app, Google Docs, Notepad, whatever you use). Create separate files for each category below. Save them as .md files (that just means plain text with a .md extension).
File 1: about-me.md
Write a few paragraphs about who you are. Keep it conversational. Include:
- Your name and what you do for work
- Your company, role, and industry
- What kinds of tasks you typically ask AI to help with
- Any personal details that matter (hobbies, interests, family context you'd want AI to know)
File 2: how-I-communicate.md
This is where you teach the AI your voice. Include:
- Your preferred tone (casual, professional, direct, warm, etc.)
- Words or phrases you love using
- Words or phrases you hate (we all have them)
- How formal or informal you want responses
- Any pet peeves about AI-generated writing
File 3: current-projects.md
A quick snapshot of what you're working on right now. This helps the AI give relevant answers without you explaining the backstory every time.
- Active projects and their status
- Key people you work with (names, roles, relationships)
- Deadlines or milestones coming up
- Tools and platforms you use daily
File 4: rules.md
Standing instructions you want every AI to follow. Things like:
- "Always use metric units"
- "Default to Python when writing code"
- "Never suggest meetings before 10am"
- "When drafting emails, always include my signature"
- "Don't use bullet points unless I ask for them"
File 5 (Optional): brand-info.md
If you have a business or personal brand:
- Your tagline, mission statement, or positioning
- Brand colors, fonts, or visual guidelines
- Target audience description
- Key differentiators
Pro tip: Write these files the way you'd talk to a new hire on their first day. Natural language works best. You don't need to be fancy or structured. Just be clear.
Step 2: Create a Folder
Pick a spot on your computer. Create a new folder. Call it something like:
my-ai-context
Put all your .md files from Step 1 inside this folder. That's your briefing folder.
Your folder should look something like this:
my-ai-context/
├── about-me.md
├── how-I-communicate.md
├── current-projects.md
├── rules.md
└── brand-info.md
Step 3: Install the Tools
You need two things installed on your computer. Both are free.
Install Node.js
Node.js is the engine that runs your MCP server. You don't need to know how to code. You just need it installed.
- Go to nodejs.org
- Download the LTS version (the one that says "Recommended")
- Run the installer. Click next through everything.
- Done.
To confirm it worked, open your terminal (Mac: Terminal app, Windows: Command Prompt) and type:
node --version
If you see a version number, you're good.
Install an AI App That Supports MCP
Right now, the main options are:
- Claude Desktop App (from Anthropic) — free, easiest setup
- Cursor (AI code editor) — if you're a developer
- Other AI tools are adding MCP support regularly. Check your favorite tool's docs.
For this guide, we'll use Claude Desktop since it's the most straightforward.
Step 4: Connect Your Folder to Your AI
This is the part that feels "technical," but it's really just editing one settings file.
For Claude Desktop (Mac)
- Open Finder
- Press
Cmd + Shift + G - Paste this path:
~/Library/Application Support/Claude/ - Look for a file called
claude_desktop_config.json. If it doesn't exist, create one. - Open it in any text editor and paste this:
{
"mcpServers": {
"my-context": {
"command": "npx",
"args": [
"-y",
"@modelcontextprotocol/server-filesystem",
"/Users/YOURUSERNAME/my-ai-context"
]
}
}
}
- Replace
YOURUSERNAMEwith your actual computer username - Replace the folder path with wherever you put your
my-ai-contextfolder - Save the file
- Restart Claude Desktop
For Claude Desktop (Windows)
- Press
Win + R - Type
%APPDATA%\Claude\and hit Enter - Same steps as above, but your path will look like:
"C:\\Users\\YOURUSERNAME\\my-ai-context"
(Note the double backslashes. Windows is picky about that.)
Step 5: Test It
Open Claude Desktop. Start a new conversation and ask:
"What do you know about me from my context files?"
If everything's connected, the AI will pull from your briefing files and tell you what it knows. It should reference your name, your work, your preferences... all the stuff you wrote down.
If it doesn't work, double check:
- Your folder path is correct (capitalization matters)
- The JSON file doesn't have any typos (missing commas break everything)
- You restarted the app after saving
Step 6: Keep It Updated
Here's the part most people skip. And it's the part that matters most.
Your context files are only useful if they reflect your current reality. Set a recurring reminder, once a month, to spend 15-20 minutes updating your files.
Things to update:
- New projects you've started
- Projects you've finished (remove them)
- New people you're working with
- Changes to your role or responsibilities
- New preferences you've discovered
- Rules that are no longer relevant
Think of it like updating your resume. Except this one actually gets read.
Why This Matters
Right now, every AI platform keeps its own version of you. ChatGPT knows one version. Claude knows another. Gemini knows nothing. And none of them talk to each other.
An MCP server flips that dynamic. Instead of each AI platform owning your context, YOU own your context. You decide what goes in. You decide what comes out. You plug it into whatever tool you want.
Switch from ChatGPT to Claude tomorrow? Your context comes with you. Start using a new AI coding tool? Plug in the same server. Your preferences, your voice, your projects... all portable.
You stop being a stranger to your own tools.
Quick Reference
| What | Where |
|---|---|
| Your briefing files |
~/my-ai-context/ folder |
| Config file (Mac) | ~/Library/Application Support/Claude/claude_desktop_config.json |
| Config file (Windows) | %APPDATA%\Claude\claude_desktop_config.json |
| Node.js download | nodejs.org |
| Claude Desktop download | claude.ai/download |
| Update frequency | Monthly (set a reminder) |
Going Further
Once you're comfortable with the basics, you can level up:
Add more folders. You can point your MCP server at multiple directories. Keep personal context in one folder and work context in another.
Make it accessible from anywhere. Tools like Tailscale let you securely access your MCP server from any device, not just the computer it lives on. This turns your local setup into something cloud-based without giving up control.
Share context with your team. Create a shared folder with company-specific context (brand guidelines, product info, team roster) and give everyone access. Now every team member's AI assistant knows the same baseline information.
Connect to other AI tools. As more tools adopt MCP, you won't need to reconfigure anything. Same folder, same files, new connection. That's the whole point.
Built once. Used everywhere. Updated monthly. That's it.
Leave a comment