Skip to main content
1

Install LongMem

Run the interactive installer:
curl -fsSL https://github.com/clouitreee/LongMem/releases/latest/download/install.sh | bash
Accept the defaults (Safe privacy mode, auto-context enabled, no compression) or customize as needed.
See the Installation guide for detailed options and troubleshooting.
2

Start the Daemon

Launch the memory daemon:
longmem start
Expected output:
✓ Daemon started (PID: 12345)
✓ Listening on port 38741
The daemon runs in the background and persists across terminal sessions. On macOS/Linux with systemd or launchd, it will auto-start on boot.
3

Verify It's Running

Check the daemon status:
longmem status
Expected output:
✓ Daemon is running (PID: 12345)
✓ Health check: OK
✓ Port 38741: accessible
✓ Memory database: 128 KB (23 observations)

$ longmem status
✓ Daemon is running (PID: 12345)
✓ Health check: OK
✓ Port 38741: accessible
4

Use With Your AI Assistant

LongMem integrates automatically with Claude Code and OpenCode. No manual configuration needed.Start a new AI session:
# Claude Code
claude-code

# OR OpenCode
opencode
Your assistant now has access to three memory tools via MCP:
ToolDescription
mem_searchSearch past sessions by keyword or semantic meaning
mem_timelineGet chronological context from recent work
mem_getRetrieve full details of a specific observation
These tools are called automatically by your AI assistant. You don’t need to invoke them manually. Just ask natural questions like:
  • “What bug did I fix yesterday?”
  • “Show me the API changes from last week”
  • “What was the error message I got earlier?”

​
Testing Memory

To verify memory is working:
1

Do some work

In your AI assistant, run a few commands:
You: Create a file called test.txt with "Hello LongMem"
AI: [creates the file]
2

Check memory stats

Exit the AI session and check stats:
longmem stats
Expected output:
Memory Statistics
─────────────────
Total observations: 3
Prompts:            1
Commands:           1
Tool outputs:       1

Database size:      4.2 KB
Oldest entry:       2026-03-04 10:23:15
Newest entry:       2026-03-04 10:23:47
3

Query memory in a new session

Start a fresh AI session and ask:
You: What file did I create in the last session?
AI: [queries mem_timeline] You created a file called test.txt

​
Where Is My Data Stored?

All LongMem data lives in ~/.longmem/: 
~/.longmem/
├── memory.db       # SQLite database with all observations
├── settings.json   # Configuration (privacy, compression, etc.)
└── logs/           # Daemon logs
Your data never leaves your machine unless you explicitly enable compression with a cloud provider (OpenRouter, OpenAI, Anthropic).

​
Useful Commands

longmem status

​
Configuration

Edit ~/.longmem/settings.json to customize behavior: 
{
  "daemon": { "port": 38741 },
  "privacy": {
    "mode": "safe",
    "redactSecrets": true,
    "customPatterns": []
  },
  "autoContext": { "enabled": true },
  "compression": {
    "enabled": false,
    "provider": "openrouter",
    "model": "meta-llama/llama-3.1-8b-instruct",
    "apiKey": ""
  }
}
Restart the daemon after config changes: 
longmem stop && longmem start

​
What Users Say

“I don’t have to repeat myself.”

“My assistant remembers yesterday’s bugs.”

“No cloud. My data stays local.”

​
Next Steps

Configuration

Customize privacy modes, compression, and advanced settings

Commands

Full CLI reference for all longmem commands

Build docs developers (and LLMs) love

Get started for freeTalk to us