05/11/2026
Give Claude Code a Memory: correct it once, and a brand new chat already knows.
Part 2 of Build an AI Employee with Claude Code.
Auto memory is notes Claude writes for itself from your corrections and preferences. It is on by default, kept per project on your computer, and its index (MEMORY.md) loads at the start of every chat, next to CLAUDE.md.
In this video:
1. Paste an email into the Claude Code panel: the draft has no sign off.
2. Correct it once: "Remember: always sign my replies with Best, Zion."
3. Start a brand new chat and paste the next email: it signs the reply by itself.
Dos and don'ts:
Do let memory keep your corrections, and check what it saved: click / in the Claude Code panel to open the command menu, then click Memory in its Customize section.
Don't keep rules in memory: rules go in CLAUDE.md. Never put passwords there.
You need the folder from Part 1, open in VS Code with the Claude Code extension.
Part 3: we teach it your world, who you are, your tone and your clients. What should it remember first?