Claude Code memory

Claude Code Memory: persistent local memory for long-running coding work.

remem gives Claude Code a searchable, auditable project memory layer for decisions, bug-fix rationale, preferences, and repo patterns that should survive fresh sessions.

Install for Claude Code Read README
  • SessionStart hooks
  • Stop summaries
  • SQLite/SQLCipher
  • MCP tools
claude - remem

$ remem install --target claude

ok Claude Code hooks configured

$ claude

SessionStart

decision release PRs need reviewThreads checked

bugfix doctor warnings can be non-install backlog

context restored before the first prompt

What remem adds

Claude Code remembers the engineering trail, not just preferences.

Use remem when the useful context is larger than a short instruction and needs search, auditability, and project scoping.

01

Captures sessions

Hooks record useful session evidence without hand-written notes.

02

Distills rationale

Background summaries preserve decisions, fixes, and lessons.

03

Searches locally

`remem search` finds project-scoped memories in the local store.

04

Injects context

Relevant memories appear at session start before work begins.

Install

Connect remem to Claude Code.

Install the binary, configure Claude Code hooks, then restart Claude Code and run a new session.

brew install majiayu000/tap/remem
remem install --target claude
remem doctor
remem search "last decision"

FAQ

Common Claude Code memory questions.

WhereLocal

Memory is stored under `~/.remem` with SQLite and SQLCipher support.

WhenSessionStart

Claude Code receives relevant project memory when a new session starts.

HowMCP + CLI

Agents can search and save memory through MCP tools or direct CLI commands.