​
Installation First

🚀 Haven't installed yet?

Start with our one-command Quick Start guide
 Already installed? Learn how to use claude-mem effectively.

​
What You Get

After installing claude-mem, you have:

​
Your First Session

1

Start Claude Code

Open any project. You’ll see claude-mem load automatically:
🧠 Claude Memory System - Active
📚 Loading relevant memories...
✨ Ready. Start coding.
2

Work Normally

Code as you always do. claude-mem captures everything in the background. No manual commands needed.
3

Save a Memory

When you want to save the conversation, type:
/compact
Or just use /clear - it saves automatically before clearing.
4

See It Work

Start a new session. Your memories load instantly:
🧠 Previous context loaded
📚 Found relevant memories from earlier sessions

​
How Memories Build Up

Each session adds to Claude’s understanding: Session 1: “I’m building a React app with TypeScript” Session 2: Claude remembers React + TypeScript Session 3: “Add authentication” → Claude knows: React + TS + auth needs Session 4: Claude suggests JWT because that’s what you used before Automatic. No commands. No configuration.

​
Understanding Memory Behavior

​
What Gets Remembered

claude-mem stores:
  • Project insights: Architecture decisions, technical requirements, and goals
  • Problem-solving patterns: How you approach and solve specific types of issues
  • Code preferences: Your coding style, preferred libraries, and patterns
  • Decision rationale: Why certain choices were made
  • Key learnings: Important discoveries and insights from your sessions

​
What Doesn’t Get Remembered

Excluded for privacy and relevance:
  • Sensitive information: API keys, passwords, or personal data
  • Verbose debugging output: Long error logs or repetitive troubleshooting
  • Off-topic conversations: Discussions unrelated to your development work
  • Transient state: Temporary variables, current session specifics

​
Memory Loading

Smart Context Loading: claude-mem loads relevant memories based on your current project. It uses the directory name to filter and prioritize the most relevant context.

​
Essential Commands

Here are the commands you’ll use most frequently:

​
File System Overview

claude-mem creates a clean directory structure: 
~/.claude-mem/
├── index/           # Compressed memory index (JSONL format)
├── archives/        # Raw conversation transcripts
├── hooks/           # Claude Code integration scripts
├── trash/           # Safe deletion recovery (Smart Trash)
└── logs/           # Operation logs for troubleshooting
Don’t edit these files manually. Use claude-mem CLI commands. Direct modification corrupts the index.

​
Next Steps

​
Quick Troubleshooting

If something doesn’t work as expected:
  1. Check Status: Run claude-mem status to identify issues
  2. Reinstall Hooks: Try claude-mem install --force
  3. Restart Claude Code: Completely restart the application
  4. Check Logs: Use claude-mem logs --error to see error details
Most issues are resolved by running claude-mem install --force and restarting Claude Code. This refreshes all integration components.
You’re ready to use persistent context in Claude Code. Every conversation builds on the last.