Quick Start

Install the package.
Approve setup explicitly.

The npm package owns its Python environment. The Python package is best installed with pipx or uv. Setup can update documented client configuration only after consent.

18+

Node.js requirement for npm

3.11–3.14

Supported Python range

Install

Use the npm package or an isolated Python application environment.

terminal — install

# Recommended interactive installer (Node.js 18+, npm 9+) $ npm install -g superlocalmemory
# Or install the Python CLI in an isolated environment $ pipx install superlocalmemory $ uv tool install superlocalmemory

The npm route requires Node.js 18+ and npm 9+. Python routes require Python 3.11–3.14. Model downloads and optional backends are separate from the core package install.

Setup

Choose your operating mode in one interactive prompt.

terminal — setup

$ slm setup # Choose your mode: # A) Local core path — no LLM provider required # B) Local Ollama — local LLM inference # C) Cloud LLM — OpenAI / Anthropic > Mode A selected. Initializing local database... ✓ Setup complete. Check slm --version for your installed package version.

Mode A does not require a model-provider API key for the core path after required dependencies and models are present. See the Modes page for the complete network boundary.

Your first memory

Store, recall, and check status — that's the full loop.

terminal — first memory

# Store your first memory $ slm remember "Alice works at Google as Staff Engineer" ✓ Stored → fact #1 · memory_confidence 0.97 · Mode A
# Recall it later $ slm recall "What does Alice do?" ↳ Staff Engineer at Google · relevance_score: 0.82 · memory_confidence: 0.97 calibration_status: uncalibrated · answer_confidence: null
# Check system status $ slm status ✓ Mode A · 1 fact indexed · cache: opt-in

Run slm --help for the installed command surface. Data-returning commands document --json where supported.

Modes

Pick your mode.

Choose a mode during slm setup. Later, switch with slm mode a, slm mode b, or slm mode c.

Local Core MODE A

No LLM needed.

Local sentence-transformer path with no LLM provider required for core memory processing. Initial model acquisition and optional integrations may use the network.

Best for: privacy-first teams, CI/CD pipelines,
any setup without GPU or API access.

$ slm mode a

→ See Mode A docs

Local Ollama MODE B

Local LLM inference.

Ollama-powered local-model path. Network and privacy behavior depend on where the Ollama endpoint runs and how models are acquired.

Use when: the configured Ollama endpoint and model meet your locality, quality, and resource requirements.
Measure RAM and latency on the selected model.

$ slm mode b

→ See Mode B docs

Cloud LLM MODE C

Provider-assisted path.

Configured provider-assisted processing. Content can transit to the provider; review its logging, retention, and transfer terms.

Use when: you accept the configured provider's data path and have measured the benefit on your workload.
Requires supported provider credentials.

$ slm mode c

→ See Mode C docs

Verification

Confirm everything is running.

Three commands to verify your install is healthy.

terminal — verification

$ slm health ✓ daemon: running · db: healthy · cache: on
$ slm status --json {"success": true, "mode": "A", "facts": 1, "cache_hits": 0}
$ slm --version superlocalmemory <installed-version>

Troubleshooting