← Back to blog
Admin 5 min read

How to Check OpenClaw Memory in MyD1

myd1sqliteopenclawsshtutorial
SQLite + OpenClaw

Check OpenClaw Memory in MyD1

Open your OpenClaw memory database directly in MyD1. Browse conversations, inspect embeddings, and query the knowledge store — all from a native GUI.

Your AI agent remembers everything it has ever learned — but without a proper tool, you will never see what is inside that database.

TL;DR

  • Local macOS/Linux: open openclaw/memory/main.sqlite directly in MyD1
  • Remote server: use MyD1's SSH bridge to connect securely — no file copying needed
  • Once connected: browse tables, run SQL, export data, or let MyD1's AI Agent write queries for you
SQLite support — including local file and SSH bridge connections — is available starting with MyD1 v2.11 and later. Make sure you are running the latest version.

What is OpenClaw's memory database?

OpenClaw stores its memory — conversations, context, embeddings, and metadata — in a SQLite database located at openclaw/memory/main.sqlite relative to the OpenClaw installation directory. This file contains everything the agent has learned and remembered across sessions.

With MyD1, you can open this database directly and browse its tables, run SQL queries, export data, and understand exactly what your agent remembers. No terminal required.

Per-platform instructions

macOS
  1. Open MyD1 and click New Connection
  2. Select SQLite as the database engine
  3. Click Browse and navigate to your OpenClaw installation directory
  4. Select the file at openclaw/memory/main.sqlite
  5. Click Connect — you will see all tables immediately
# Typical path on macOS
~/openclaw/memory/main.sqlite

# Or if installed via Homebrew
/usr/local/share/openclaw/memory/main.sqlite

Tip: You can also drag and drop the .sqlite file directly onto the MyD1 window to open it instantly.

🐧
Linux
  1. Locate your OpenClaw memory database on the Linux machine
  2. If running MyD1 on the same machine, use the local SQLite connection (same steps as macOS)
  3. If accessing from a remote Mac, use the SSH bridge method described below
# Common paths on Linux
~/openclaw/memory/main.sqlite
/home/<user>/openclaw/memory/main.sqlite

# Check if the file exists
ls -la ~/openclaw/memory/main.sqlite

SSH bridge — read a remote SQLite database securely

If your OpenClaw instance runs on a remote server — for example an Ubuntu VPS — you do not need to copy the database file. MyD1's SSH tunnel support lets you connect securely to the remote machine and read the SQLite file directly, fully encrypted over SSH.

MyD1 (your Mac) → SSH Tunnel (encrypted) → Ubuntu Server → main.sqlite

Step-by-step: Ubuntu server

  1. Open MyD1 and click New Connection
  2. Select SQLite as the database engine
  3. Enable SSH Tunnel in the connection form
  4. Enter your server details: 
    # SSH connection
Host: your-server.example.com
Port: 22
Username: ubuntu
Auth: SSH Key (recommended) or Password

# Database path on the remote server
File: /home/ubuntu/openclaw/memory/main.sqlite
  5. If using SSH key auth, MyD1 auto-detects keys in ~/.ssh/ — select your key or let it use the default id_rsa / id_ed25519
  6. Click Connect — MyD1 establishes the SSH tunnel and opens the remote database as if it were local

The connection is fully encrypted end-to-end. Your database contents never leave the SSH tunnel. No ports need to be opened on the server beyond SSH (port 22).

Example: hardened Ubuntu setup

# On your Ubuntu server — lock down SSH
sudo ufw allow 22/tcp
sudo ufw enable

# Ensure key-only auth (disable password login)
sudo sed -i 's/PasswordAuthentication yes/PasswordAuthentication no/' /etc/ssh/sshd_config
sudo systemctl restart sshd

# Make sure the OpenClaw user can read the database
chmod 644 /home/ubuntu/openclaw/memory/main.sqlite

If the SQLite file is actively being written to by OpenClaw, consider connecting in read-only mode or stopping the OpenClaw process first to avoid WAL conflicts.

What you can do once connected

With the memory database open in MyD1, you get the full toolkit:

  • Browse tables — see conversations, memory entries, embeddings, and metadata at a glance
  • Run SQL queries — search memory by keyword, filter by date, count entries
  • Export data — dump the full memory to CSV, SQL, or TSV for analysis
  • Schema inspection — see column types, constraints, and indexes
  • AI Agent — ask questions about the memory schema in plain English and MyD1's AI Agent writes the SQL for you
-- Example: find the 10 most recent memory entries
SELECT * FROM memory
ORDER BY created_at DESC
LIMIT 10;

-- Example: count entries by type
SELECT type, COUNT(*) AS total
FROM memory
GROUP BY type
ORDER BY total DESC;