SuperLocalMemory V3 installs via npm, pip, or git clone. All three methods give you the same product — choose whichever fits your workflow.

No desktop app (DMG/EXE) for V3. V3 is a CLI + MCP server, not a GUI application. The V2 desktop installers are deprecated. Use slm dashboard for the web UI.

Python 3.11+ is required for the V3 engine. Node.js is only needed if you install via npm.

One command installs everything — CLI, Python dependencies, and MCP server.

npm install -g superlocalmemory

This automatically:

• Installs the V3 engine and CLI (slm command)
• Auto-installs Python dependencies (numpy, scipy, networkx, sentence-transformers, torch)
• Creates data directory at ~/.superlocalmemory/
• Auto-installs Claude Code hooks (v3.3.6+) — memory lifecycle is fully automatic
• Detects V2 installations and guides migration

That's it. Open Claude Code and memory just works. No slm setup or slm init needed for auto-memory.

For optional configuration:

slm setup # Interactive wizard — choose Mode A/B/C, configure provider
slm warmup # Pre-download embedding model (~500MB, one-time)

slm warmup is optional. If you skip it, the model downloads automatically on your first slm remember or slm recall.

Don't want auto-hooks? Run slm hooks remove to opt out. Re-enable anytime with slm hooks install.

slm status

You should see:

SuperLocalMemory V3
 Mode: A
 Provider: none
 Base dir: /home/you/.superlocalmemory
 Database: /home/you/.superlocalmemory/memory.db

pip install superlocalmemory

Then run:

slm setup
slm warmup # Optional — pre-download embedding model
slm status # Verify

git clone https://github.com/qualixar/superlocalmemory.git
cd superlocalmemory
pip install -e .

Then:

slm setup
slm warmup
slm status

Total disk footprint: ~750MB after first use (mostly PyTorch + embedding model).

RAM usage: ~500-800MB peak during embedding model load, ~20-50MB steady state. CPU-only — no GPU required.

If any dependency fails during install, the installer prints the exact pip install command to fix it. BM25 keyword search works even without embeddings — you're never fully blocked.

npm install -g superlocalmemory
slm setup

Works out of the box. Python 3.11+ is included with Homebrew (brew install python@3.12) or available from python.org.

npm install -g superlocalmemory
slm setup

Ensure Python 3.11+ is installed: sudo apt install python3.11 (Ubuntu) or sudo dnf install python3.11 (Fedora).

npm install -g superlocalmemory
slm setup

Requires Python 3.11+ from python.org. Add Python to PATH during installation.

After installing, connect to your AI IDE:

{
 "mcpServers": {
 "superlocalmemory": {
 "command": "slm",
 "args": ["mcp"]
 }
 }
}

Or auto-configure all detected IDEs:

slm connect # Configure all detected IDEs
slm connect --list # See which IDEs are configured

See IDE Setup for per-IDE instructions.

If you have V2 (2.8.6 or earlier) installed:

npm install -g superlocalmemory # Installs V3 alongside V2
slm migrate # Migrates V2 data to V3 schema

V3 is a complete architectural reinvention — new mathematical engine, new retrieval pipeline, new storage schema. Your existing data is preserved. A backup is created automatically before migration.

See Migration from V2 for the full guide.

• npm install: Make sure npm global bin is in your PATH. Run npm bin -g to find the location.
• pip install: Make sure Python scripts directory is in your PATH.

• Ensure Python 3.11+ is the default: python3 --version
• Reinstall: pip install --force-reinstall superlocalmemory

• Check internet connection
• Try manual warmup: slm warmup
• If behind a proxy, set HTTP_PROXY and HTTPS_PROXY environment variables

• Use npm install -g superlocalmemory (not sudo)
• If npm global directory needs permissions: npm config set prefix ~/.npm-global and add ~/.npm-global/bin to PATH

• Quick Start Tutorial — Your first memory in 2 minutes
• Modes Explained — Choose between A (zero-cloud), B (local Ollama), C (full power)
• CLI Reference — All 14 commands with examples

Part of Qualixar | Created by Varun Pratap Bhardwaj