Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type @ followed by the MCP server name and your instructions, e.g., "@Smara Memory MCP Server remember that I prefer using TypeScript and Tailwind CSS for my web projects"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
MCP server for the Smara Memory API — give any AI app persistent memory with Temporal Memory Scoring™.
Add to your Claude Desktop config (~/Library/Application Support/Claude/claude_desktop_config.json):
{
"mcpServers": {
"smara": {
"command": "npx",
"args": ["-y", "@smara/mcp-server"],
"env": {
"SMARA_API_KEY": "your-api-key-here"
}
}
}
}Get a free API key at smara.io (10,000 memories, no credit card).
Related MCP server: Mnemex
Tool | Description |
| Store a fact about a user with importance scoring |
| Semantic search with Temporal Memory Scoring™ |
| Pre-formatted context string for LLM system prompts |
| Delete a specific memory |
| Check plan limits and memory count |
Smara uses Temporal Memory Scoring™ — a proprietary ranking system that makes AI memory work like human recall. Memories naturally fade over time, modulated by importance and access patterns. Recent, critical memories surface first. Stale, trivial ones fade. Contradictions are auto-detected and resolved.
Claude Desktop
Claude Code
Cursor
Windsurf
VS Code (GitHub Copilot)
Any MCP-compatible client
Variable | Required | Default |
| Yes | — |
| No |
|
MIT
Delete a specific memory by ID. Use when a user asks to forget something.
Inputs: id (required).
Check current memory usage — plan, limits, and how many memories are stored. Useful for monitoring quota.
Inputs: no named parameters.
Retrieve a pre-formatted context string for a user, ready to inject into an LLM system prompt. Ranked by Temporal Memory Scoring. Can be called without a query to get the most important recent memories. When a team is configured, includes team memories automatically.
Inputs: user_id (required), q, top_n, namespace, +2 more.
List stored memories for a user with their visibility status (private or team). Use when the user wants to see what's remembered, review team vs private memories, or find a memory to update/delete.
Inputs: user_id (required), limit, offset, namespace, +2 more.
Semantic search across stored memories for a user. Ranked by Temporal Memory Scoring — balances semantic relevance with memory freshness and importance. When a team is configured, returns both private and team memories by default.
Inputs: user_id (required), q (required), limit, namespace, +2 more.
Store a fact or preference about a user. Smara handles deduplication and contradiction detection automatically. Use importance 0.1-0.3 for trivia, 0.5 for general facts, 0.7-1.0 for critical preferences.
Inputs: user_id (required), fact (required), importance, namespace, +2 more.
Update a memory's visibility (private/team) or importance. Use when a user wants to make a memory private, share it with the team, or change its importance. Find the memory ID via search_memories or list_memories first.
Inputs: id (required), visibility, importance.
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
Scored across 7 tools
Each tool has a clearly distinct purpose with no overlap. For example, 'list_memories' shows stored memories, 'search_memories' performs semantic search, 'store_memory' creates new memories, 'update_memory' modifies existing ones, 'delete_memory' removes memories, 'get_user_context' retrieves formatted context, and 'get_usage' monitors system usage. The descriptions reinforce these unique roles, eliminating any ambiguity.
All tool names follow a consistent verb_noun pattern using snake_case, such as 'list_memories', 'search_memories', 'store_memory', 'update_memory', 'delete_memory', 'get_user_context', and 'get_usage'. This uniformity makes the tool set predictable and easy to navigate, with no deviations in naming conventions.
With 7 tools, this server is well-scoped for a memory management system. The count is appropriate as it covers essential operations like CRUD (create, read, update, delete), search, context retrieval, and usage monitoring without being excessive or insufficient. Each tool serves a clear, necessary function in the domain.
The tool set provides complete coverage for memory management, including storing, listing, searching, updating, deleting, retrieving context, and monitoring usage. There are no obvious gaps; it supports full lifecycle management from creation to deletion, with additional utilities for context and system health, ensuring agents can handle all typical workflows without dead ends.
Persistent memory and knowledge management for AI agents with semantic search and 50+ tools.
Universal memory for AI agents and tools. Save, organize and search context anywhere.
Hosted persistent memory with semantic search, importance and TTL for AI agents.
Persistent memory for AI agents: save, search and connect context across tools and sessions.
