Claude Code has a /memory command that appends facts, preferences, and constraints to a CLAUDE.md file — either global (~/.claude/CLAUDE.md) or project-level. It loads automatically at session start. No prompt engineering needed.

This is exactly the "long-term knowledge" layer from my context management system, automated by the tool. And it works well for what it does: persistent project identity, tech stack preferences, coding conventions. The stuff that rarely changes.

But it's one layer out of five. /memory doesn't track what you've built (PROGRESS.md), why you made specific decisions (DECISIONS.md), how to hand off between sessions (HANDOVER.md), or what the implementation details look like (TECHNICAL.md). It also appends unstructured bullet points rather than maintaining deliberate sections — stakeholder profile, engineering standards, and a note about your preferred indentation all live in the same flat list.

The tooling is automating the easiest context layer first, because persistent facts are the most obvious problem. The harder ones — state tracking, decision logging, session continuity, and role separation across agent teams — are still manual. /memory makes the entry point easier without raising the ceiling.