How Claude Remember Your Project

Most people use Claude Code wrong because they keep re-explaining their project every session. They don't have to. CLAUDE.md and auto memory exist exactly for this — but most people don't know how they actually work. 🎯 In 15 minutes, the complete guide to making Claude Code remember your project — from CLAUDE.md basics to path-scoped rules, auto memory, and the troubleshooting tricks that save you hours. This is Episode 4 of the Claude Code series. Episode 1 covered architecture, Episode 2 extensions, Episode 3 the context window. Today — the memory systems. ⚡ What you'll learn: ✅ The two memory systems built into Claude Code (and when to use which) ✅ Where CLAUDE.md actually lives — all FOUR scopes explained ✅ When to add to CLAUDE.md (the 4-trigger rule) ✅ How /init bootstraps your CLAUDE.md from scratch ✅ The 4 rules for instructions Claude actually follows ✅ Vague vs specific — 3 examples that prove the difference ✅ @path imports for splitting big files ✅ How CLAUDE.md files actually load (directory tree walk) ✅ .claude/rules/ — modular conventions that scale ✅ ⭐ Path-scoped rules — the killer feature most people miss ✅ Auto memory — what Claude saves automatically ✅ The /memory command (use this DAILY) ✅ Troubleshooting: 4 problems, 4 fixes The path-scoped rules section at minute 6 will change how you organise every Claude Code project from now on.