The 200–500 line rule

Build-along · ~8 min at your own keyboard

Every CLAUDE.md line gets injected into every conversation. A 10K-token file taxes each session before you type. Keep it 200–500 lines, put the hardest guardrails at the very top (models remember beginnings and endings, not middles), and prune it like a stale listing. The way out is progressive disclosure: keep the always-needed rules in the file, and for the rest leave a one-line pointer ('pricing rules: see pricing.md') so Claude reads the detail only when the task needs it.

Where these ideas come from: Progressive disclosure (AI Hero dictionary). Paraphrased here.

Do this

0/4

Share what happened

Paste your before/after line counts, the 3 rules you promoted to the top, and the /context delta.