Developing Your Own Operating Rules

Porting Cursor Rules To Your Life

Using tools like Cursor has really opened my eyes toward the ways that life is programmable.

The parallels are striking. At our core, we are machines that react to all types of inputs. Those inputs don't just appear when we're at work or living life. They appear in our interpretation of what happened in our daily lives. And the older I get, the more I recognize that interpretation and our inner monologues are silently pulling the strings.

In Cursor (or any other AI-powered coding editor), it's essential to create a list of rules. The rules are essentially what you think they are. They tell the LLM what rules and practices to follow when working on your project.

In my projects, I typically have several rules files, including an "Always" file, a "Components" file, a "Project Context" file, and a "Typescript" file.

Each file is slightly different, but the Always file is the most relevant here. It provides a list of rules that I've developed (or borrowed) that put me in the best position to build the product. This includes everything from telling it to update the changelog after every feature is complete to confirming that I'm not revealing secrets or keys before pushing any commit.

Like I've said before, the value isn't just that the LLM will respect these rules when working on my project. Arguably, the greater value is creating the rules themselves. It's the reflective process that's required to make a file like this.

And these rules aren't necessarily set in stone. As I come across issues, I add solutions to future iterations of the Always file.

Because I've found it so helpful, I started thinking of other ways that I could use an Always file in my everyday life. It ended up becoming a quick exercise with ChatGPT. Because it has context about my professional background, goals, and strengths, I figured that it could deliver a good starting point in creating my Always file.

Ultimately, it came back with some pretty good ideas! Some of them really hit close to home. This first draft of my Always file included the following:

DEFAULT BEHAVIORS:

• Bias to shipping over optimizing.
• Operate in small, composable bets.
• Assume momentum compounds faster than insight.

MENTAL OPERATING SYSTEM

• Zoom out before emotional decisions.
• Separate identity from current output.
• Assume feelings are signals, not commands.

Granted, it's not perfect. But at the same time, I knew it wouldn't be. I wanted an outside perspective as I look inward and try to reach my, admittedly, high goals. After I created v1 of my Always file, I started to look at it once every morning before work (with occasional iterations, as necessary).

Like most things I write about, I encourage you to try this for yourself. If you do, let me know how it goes! You may be surprised by what you find.

Prompt of the Week

No prompt of the week today because of what I described above. I challenge you to create an Always file for your own life! Let me know how it goes.