Skip to content

Quick Start Tutorial

Jump to bottom
Varun Pratap Bhardwaj edited this page Feb 7, 2026 · 13 revisions

Quick Start Tutorial

Your first memory in 2 minutes - Get started with SuperLocalMemory V2's intelligent local memory system for AI coding assistants.

Prerequisites

Before starting, ensure you have:

  • Python 3.8 or higher installed (python3 --version)
  • Git installed (for cloning repository)
  • 5 minutes of time

Already installed? Skip to Your First Memory.

Installation (60 seconds)

Mac/Linux

# Clone repository
git clone https://github.com/varun369/SuperLocalMemoryV2.git
cd SuperLocalMemoryV2

# Install
./install.sh

Windows (PowerShell)

git clone https://github.com/varun369/SuperLocalMemoryV2.git
cd SuperLocalMemoryV2
.\install.ps1

What happens during installation:

  1. Creates ~/.claude-memory/ directory
  2. Initializes SQLite database
  3. Auto-detects installed IDEs (Cursor, Windsurf, Claude Desktop, VS Code)
  4. Configures MCP server for detected tools
  5. Installs CLI wrapper (slm command)
  6. Sets up shell completions (bash/zsh)

Verify installation:

slm status

Expected output:

✓ Database: OK (0 memories)
✓ Knowledge Graph: Ready
✓ Pattern Learning: Ready
✓ Current Profile: default

For detailed installation troubleshooting, see the Installation Guide.

Your First Memory

Step 1: Save Your First Memory

slm remember "We use FastAPI for REST APIs in this project"

Output:

Memory added with ID: 1

✅ Memory saved successfully

Next steps:
  • Use `slm recall <query>` to search this memory
  • Use `slm list` to see recent memories

What just happened:

  • Content saved to local SQLite database (~/.claude-memory/memory.db)
  • TF-IDF vectors generated for semantic search
  • Entities extracted ("FastAPI", "REST APIs")
  • Pattern learning analyzed your preference for FastAPI
  • Full-text search index updated

Step 2: Add More Context

slm remember "JWT tokens expire after 24 hours" --tags security,auth,jwt

With tags for better organization:

  • Helps filtering
  • Improves search relevance
  • Enables tag-based queries

Step 3: Add Project-Specific Memory

slm remember "Database uses PostgreSQL 15 with UUID primary keys" --project myapp --tags database,postgresql

Project isolation prevents context bleeding between different codebases.

Step 4: Add Important Information

slm remember "CRITICAL: Production deploys require approval from @lead" --importance 10 --tags deployment,production

Importance levels (1-10):

  • 1-3: Low priority (notes, ideas)
  • 4-6: Normal (coding patterns, decisions)
  • 7-9: High priority (critical info, warnings)
  • 10: Critical (blockers, security issues)

Search Your Memories

Basic Search

slm recall "FastAPI"

Output:

🔍 Search Results (1 found)

[ID: 1] Score: 0.95
We use FastAPI for REST APIs in this project
Tags: -
Project: default
Created: 2026-02-07 14:23

Three search methods working simultaneously:

  1. Semantic Search (TF-IDF) - Finds conceptually similar content
  2. Knowledge Graph - Discovers related memories via graph edges
  3. Full-Text Search (FTS5) - Exact keyword matching

Advanced Search

# Limit results
slm recall "authentication" --limit 3

# Filter by relevance score
slm recall "React hooks" --min-score 0.7

# Filter by tags
slm recall "database" --tags postgresql,performance

# Filter by project
slm recall "API design" --project myapp

See the CLI Cheatsheet for all search options.

List Recent Memories

slm list --limit 10

Output:

📋 Recent Memories (3 total)

[ID: 3] CRITICAL: Production deploys require approval from @lead
Tags: deployment, production
Project: default
Created: 2026-02-07 14:25

[ID: 2] JWT tokens expire after 24 hours
Tags: security, auth, jwt
Project: default
Created: 2026-02-07 14:24

[ID: 1] We use FastAPI for REST APIs in this project
Tags: -
Project: default
Created: 2026-02-07 14:23

Build Knowledge Graph

After adding several memories, build the knowledge graph to discover relationships:

slm build-graph

Output:

🔄 Building Knowledge Graph...

Phase 1: Entity Extraction
  Scanning 3 memories...
  Extracted 15 unique entities
  Created 15 graph nodes
  ✓ Complete (0.1s)

Phase 2: Relationship Discovery
  Computing similarity scores...
  Created 8 edges (relationships)
  Avg edges per node: 0.5
  ✓ Complete (0.2s)

✅ Knowledge graph built successfully!

Graph Statistics:
  Nodes: 15
  Edges: 8
  Density: 0.37%

Why build the graph:

  • Improves search quality
  • Discovers hidden relationships
  • Enables graph-enhanced recall
  • Creates topic clusters

Learn more in the Knowledge Graph Guide.

Check System Status

slm status

Output:

📊 SuperLocalMemory V2 Status

Database:
  Total memories: 3
  Database size: 12 KB
  Location: /Users/username/.claude-memory/memory.db

Knowledge Graph:
  Nodes: 15
  Edges: 8
  Clusters: 0 (run with --clustering to enable)

Pattern Learning:
  Learned patterns: 2
  Confidence threshold: 0.5

Current Profile: default

Use Across Different Tools

SuperLocalMemory works across 11+ IDEs and tools. All use the same local database - no data duplication.

In Cursor (MCP Integration)

After installation, Cursor automatically detects SuperLocalMemory.

Usage in Cursor:

You: "Remember that we use FastAPI for REST APIs"
AI: [Automatically invokes memory tools]
✓ Memory saved

You: "What did we decide about APIs?"
AI: [Searches memories]
Found: "We use FastAPI for REST APIs in this project"

In Claude Code (Skills)

# Use skills directly
/slm-remember "content"
/slm-recall "query"
/slm-list-recent
/slm-status

In VS Code with Continue.dev

# Slash commands
/slm-remember "content"
/slm-recall "query"

In Any Terminal

# Universal CLI works everywhere
slm remember "content"
slm recall "query"
slm list
slm status

See Multi-Profile Workflows for advanced usage.

Next Steps

Congratulations! You've completed the quick start tutorial.

Explore Advanced Features

  1. Multi-Profile Workflows - Separate work/personal contexts
  2. Knowledge Graph Guide - Deep dive into graph features
  3. Pattern Learning - How SuperLocalMemory learns your style
  4. CLI Cheatsheet - Complete command reference
  5. Python API - Programmatic access

Common Use Cases

  • Daily Standups: slm recall "decided" --limit 5
  • Code Reviews: slm remember "Code review feedback: Use async/await consistently" --tags codereview
  • Bug Tracking: slm remember "Bug: JWT tokens expire too fast" --tags bug --importance 8
  • Architecture Decisions: slm remember "ADR: Use microservices over monolith" --tags architecture --importance 9

Get Help

Screenshots (Coming Soon)

  • Installation process
  • First memory saved
  • Search results with scores
  • Knowledge graph visualization
  • IDE integrations (Cursor, Windsurf, Claude Desktop)

Troubleshooting

"slm: command not found"

Solution:

# Restart your shell
exec $SHELL

# Or manually add to PATH
echo 'export PATH="$HOME/.claude-memory/bin:$PATH"' >> ~/.bashrc
source ~/.bashrc

"Database not found"

Solution:

# Verify installation
ls -la ~/.claude-memory/memory.db

# If missing, reinstall
cd SuperLocalMemoryV2
./install.sh

"No memories found" after saving

Solution:

# Check database
sqlite3 ~/.claude-memory/memory.db "SELECT COUNT(*) FROM memories;"

# If 0, check write permissions
chmod 755 ~/.claude-memory/
chmod 644 ~/.claude-memory/memory.db

For more troubleshooting, see the Configuration guide.

Summary

You've learned how to:

  • ✅ Install SuperLocalMemory V2
  • ✅ Save your first memory
  • ✅ Search memories semantically
  • ✅ List recent memories
  • ✅ Build knowledge graph
  • ✅ Check system status
  • ✅ Use across different tools

100% local. 100% private. 100% yours.

Created by Varun Pratap Bhardwaj Solution Architect • SuperLocalMemory V2

GitHub • Issues • Wiki

SuperLocalMemory V3 — Your AI Finally Remembers You. 100% local. 100% private. 100% free.

Part of Qualixar | Created by Varun Pratap Bhardwaj | GitHub

SuperLocalMemory V3

Getting Started

Reference

Architecture

Enterprise

Release Notes

V2 Documentation

Clone this wiki locally