> For the complete documentation index, see [llms.txt](https://guide.ivanzakutnii.com/the-hard-way-to-swe-excellence/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://guide.ivanzakutnii.com/the-hard-way-to-swe-excellence/learning-path/developing-sustainable-programming-skills.md).

# Developing Sustainable Programming Skills

By now, you should understand your first programming language reasonably well, know the basic tools around it, and have started using Git.

Now build fluency: practise until common operations in the language no longer consume all of your attention and you can focus on the problem.

Choose one problem-solving resource with difficulty levels. Start slightly above the elementary tasks you can already solve comfortably.

If you are using Codewars, working toward 6-5 kyu problems is a useful rough target. Rank is only a proxy; the real goal is to recognize recurring problem shapes and express a correct solution without flailing through syntax.

> Give every problem at least 15 focused minutes before looking for outside help. Longer is useful while you are still making progress.

When you use an LLM, ask for the smallest intervention that gets you moving: a question about your reasoning, a hint, a counterexample, or a review of your attempted solution. Do not turn the practice set into a prompt-writing exercise.

Keep your solutions in a well-organized Git repository so you can revisit them. Publish the repository only when the exercise platform permits public solutions.

Do not get carried away with the counter. One hundred honestly completed tasks may teach more than five hundred copied ones, but the number is not the acceptance criterion.

If you did not choose Codewars, increase the difficulty until problems require decomposition, edge-case analysis, and a deliberate choice of data structure.

Move on when you can solve a representative mix of problems, test edge cases, compare two possible solutions, and explain their time and space costs. Clicking through a large pile of nearly identical tasks will not build that ability.