Let the Claude Mac app read your Apple Notes

Hello Reader,

I switched from ChatGPT to Claude when Anthropic stood its ground against the Pentagon.

It started as an ethics call. It became a Mac call.

Why This Matters

I still run macOS 14 Sonoma on my trusty 2020 Intel iMac.

The ChatGPT and Gemini desktop apps will not install on it.

Both require Apple Silicon and at least macOS 15 Sequoia.

The Claude desktop app installed fine.

And it puts Claude Chat, Code, and Cowork in one window, so I stopped juggling separate tools.

But the real unlock came later.

Connect Your AI Assistant to Your Mac-Only Apps

The Claude desktop app supports MCP connectors.

MCP stands for Model Context Protocol.

It is a standard way to let an AI assistant read from and write to other apps.

Some connectors are built and approved by Anthropic.

Others are open-source projects you add yourself from GitHub.

1. Reach into apps that only exist on a Mac: Connectors let Claude work with Apple Notes, NotePlan, Timing, and many more Mac-only apps.
2. Ask in plain language: Once the Timing MCP connector is active, I can ask things like “how much time did I spend in each app today?” and the assistant reads my actual data to answer. I built and published my own Timing MCP server for exactly this. It is open-source and only accesses the Timing database in read-only mode.
3. Keep it private by design: A local connector runs on your machine, so your data is not sent anywhere just by running it. With a cloud assistant like Claude, only the specific query result leaves your Mac. For fully local use, you can point the same connector at a local AI tool instead.

Timing is the time-tracking app behind that last example. You can get it on its own, or as part of a Setapp subscription, where new subscribers get a 30-day free trial instead of the usual 7 using my affiliate link.

Pro Tip

Adding open-source connectors from GitHub raises one real question: can you trust the code?

I built an assistant to answer that for me. I call it my MCP Security Evaluator.

I point it at any GitHub repository, and it returns a full report:

• scores for security, privacy, and dependencies
• the main risks flagged, and suggested fixes
• it even surfaces alternative servers

A manual review like that used to take me an hour or more per server.

Now I get a verdict in a couple of minutes.

I do not install anyone else’s MCP server until this assistant gives it a passing grade or a list of fixes that Claude Code could tackle for me.

I have packaged this assistant as a downloadable skill you can use yourself: macpreneur.com/mcp-evaluator.

This week’s Macpreneur episode is where this all gets real.

The MCP Security Evaluator is just one of the many assistants I walk through.

I cover the full stack I run my business on, how I structure each one, and how you can build your own.

I even hand you a head start.

My free AI Co-C-Suite Implementation Kit gives you a co-CEO assistant and a guide to put it to work today.

👉 Access the complete show notes and video version of episode #174​

Cheers,

Damien

Affiliate disclosure: Some of the links in this email are special. Why? Because if you click on them and make a purchase, I earn a small commission at no extra cost to you. It’s like a virtual high-five for recommending stuff I love! So, thank you for supporting me and the Macpreneur podcast! Remember, I only promote products that I genuinely believe in.