What Works and What Doesn't in CLAUDE.md
Writing "Write clean code" in CLAUDE.md changes nothing. During the process of building 10 personal apps in three months, I rewrote CLAUDE.md many times. Since it became clear what worked and what didn't, I will outline that distinction. What Doesn't Work Giving instructions with adjectives - Write clean code - Design for maintainability - Handle errors appropriately - Consider performance None…
Writing clean code in CLAUDE.md changes nothing. During the process of building personal apps, I rewrote CLAUDE.md many times. Clear instructions that can be verified, such as not using pure black or white colors, do not use gradient backgrounds, and limit to three types of motion per screen. These rules can be checked with grep.
Verifying that code is clean is difficult, so it does not work well. Limiting permissions by explicitly stating what cannot be done is more effective than suggesting what should be done. Providing an exit, such as "stop implementation and report it," helps the model retreat when it hits constraints. Writing exceptions in advance avoids inconsistent judgments. Keeping the file short and focusing on verifiable rules works best.
Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
