Persistent memory for
AI assistants
Give your AI assistant a memory that persists across conversations. Works with Claude Code, Cursor, Windsurf, and any MCP-compatible client.
{
"mcpServers": {
"memory": {
"command": "npx",
"args": ["@lakehouse42/memory-mcp"]
}
}
}Everything your AI needs to remember
Remember Everything
Store facts, preferences, decisions, and context. Your AI assistant remembers what matters.
Semantic Recall
Find relevant memories by meaning, not just keywords. BGE-M3 embeddings power 91% recall accuracy.
Instant Access
Sub-50ms retrieval. Memories surface as fast as your AI can think.
Private & Secure
Your data stays yours. Local mode works offline, cloud mode is encrypted end-to-end.
Deduplication
3-tier semantic deduplication ensures no duplicate memories, keeping context clean.
Knowledge Graph
Entities and relationships are automatically extracted and connected.
Simple, powerful API
remember
Store a memory for later recall
remember("The user prefers dark mode")recall
Search memories by semantic similarity
recall("user preferences")forget
Delete outdated information
forget("memory-id-123")list_memories
List recent memories
list_memories()memory_status
Check system status
memory_status()Start free, upgrade when ready
Local mode works immediately with no signup. Connect to LH42 for full semantic search and enterprise features.
| Feature | Free (Local) | Lakehouse |
|---|---|---|
| Semantic search | - | |
| Local storage | ||
| 3-tier deduplication | - | |
| Knowledge graph | - | |
| Multi-device sync | - | |
| Temporal history | - | |
| Importance decay | - | |
| API access |
Three steps to persistent memory
Add to config
Add the MCP server to your Claude Code or Cursor configuration. One line, done.
Your AI remembers
As you work, your AI naturally stores important facts, preferences, and context.
Context carries over
Start a new conversation and your AI already knows your preferences and history.
Ready to give your AI memory?
Start with local storage for free, no signup required. Upgrade to LH42 for full semantic search.