Claude Code Memory
Claude Code project memory without turning CLAUDE.md into a wiki.
Memory works with Claude Code by keeping behavioral guidance in `CLAUDE.md` and moving evolving project knowledge into local, typed, reviewable memory.
What this solves.
Small CLAUDE.md Keep instructions stable and project knowledge retrievable.
`memory setup` writes compact Claude Code guidance, while Memory stores product intent, decisions, conventions, and traps where agents can retrieve them by task.
Session continuity Carry durable context across Claude Code sessions.
Claude Code can start from `memory load` instead of asking you to repeat repository context after every new session or context reset.
Optional MCP Use the CLI first, then add MCP when Claude exposes it.
The default workflow stays simple: `memory load`, work, `memory remember --stdin`, and inspect with `memory diff` or `memory view`.
Related Memory guides.
Continue with adjacent project-memory workflows for AI coding agents, AI coding assistants, and MCP-capable tools.