About

About 20 years ago, as a young developer, I broke production. The rollback was slow, and for a few hours I was sure the company would close because of me. Looking back, the impact was barely worth mentioning. At the time, I was devastated.

A few days later, the team went out for drinks, and someone gave me advice I have used ever since: “Anticipate failure. Write down what might go wrong. Then, hope you never have to think about it again.” I no longer remember who said it.

It took a few more incidents and near misses before the advice stuck. I started listing what could go wrong before any change I cared about, and sometimes the list changed the plan. I cannot point to the moment it became a habit.

That list eventually became a kata called the Pre-Mortem Minute. Before you commit to a technical approach, you write “This has failed in production. Why?”, list 3 specific reasons, and check whether your plan addresses the most plausible one.

Years later, I replayed that first incident with the kata in mind. The most plausible risk was a schema migration that ran on deploy with no tested way to reverse it. Running it on its own before the code change would have removed that risk.

I cannot be sure I would have caught it. I only know I never gave myself the chance.

Habits like that one matter most after the move into a lead role, because that is when the feedback that rewarded earlier habits goes quiet. Advice gets vaguer at the same time. A goal like “communicate more clearly” gives you nothing to do at 11:42 on a Tuesday.

Nothing pushed me to put my habits on paper until March 2026, when we started training the next group of tech leads at Epignosis. They had to become something I could hand over, so I extracted the katas from my internal presentation and published them in the open. Anyone can clone, fork, or argue with them.

Each kata takes 10 to 15 minutes and needs no team, coach, or permission. You do not need to be a tech lead yet to practice one.

Stay with one kata for a week. Keep a short note as you go, with 3 questions:

• Did the moment arise?
• What did I do?
• What would I change?

If the moment never came up, that is also worth knowing.

I make no promises about results. I have seen these katas work, sometimes for me and sometimes for people I passed them to over the years. I have also seen versions that worked for me fail to land for someone else, and I have abandoned some myself.

The best contribution is to pick a kata and try it out. A week of practice shows where a kata could be clearer and which moments the catalog does not cover yet.

When you want to turn what you noticed into a change, the katas live in a public GitHub repository, and contributions of every size are welcome:

• New katas for engineers preparing for a lead role
• Clearer versions of existing katas that keep the same behavior
• Fixes to examples and documentation

Every kata started as a habit. The next one could be yours.