@memsolus/mcp

Persistent memory for AI agents. One line to install, zero config to remember everything.

Give your AI agents the ability to store, recall, and reason over long-term memory — preferences, facts, decisions, relationships, and structured knowledge — across conversations and sessions.

Built on the Model Context Protocol, works with Claude, GPT, and any MCP-compatible client.

Why Memsolus?

• Semantic search — Find memories by meaning, not just keywords. Hybrid search combines both for best results.
• Knowledge graph — Entities and relationships are automatically extracted. Ask "Who works on Project X?" and get structured answers.
• Auto-consolidation — Raw memories are processed into structured knowledge profiles, versioned and categorized.
• Memory pools — Shared memory spaces for teams and multi-agent collaboration.
• Priority-aware — Mark critical rules as HIGH priority. They rank higher in search and resist pruning.
• Multi-tenant — Isolate context per user, agent, or workspace. Fine-grained scoping built in.

Setup

Get your API key at app.memsolus.com/api-keys, then add the server to your tool:

Claude Desktop

Add to ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows):

{
  "mcpServers": {
    "memsolus": {
      "command": "npx",
      "args": ["-y", "@memsolus/mcp"],
      "env": {
        "MEMSOLUS_API_KEY": "mk_live_..."
      }
    }
  }
}

Claude Code

claude mcp add memsolus -- npx -y @memsolus/mcp

Then set your key in .claude/settings.local.json:

{
  "env": {
    "MEMSOLUS_API_KEY": "mk_live_..."
  }
}

Cursor

Add to .cursor/mcp.json in your project root:

{
  "mcpServers": {
    "memsolus": {
      "command": "npx",
      "args": ["-y", "@memsolus/mcp"],
      "env": {
        "MEMSOLUS_API_KEY": "mk_live_..."
      }
    }
  }
}

Windsurf

Add to ~/.codeium/windsurf/mcp_config.json:

{
  "mcpServers": {
    "memsolus": {
      "command": "npx",
      "args": ["-y", "@memsolus/mcp"],
      "env": {
        "MEMSOLUS_API_KEY": "mk_live_..."
      }
    }
  }
}

VS Code (Copilot)

Add to .vscode/mcp.json in your project:

{
  "servers": {
    "memsolus": {
      "command": "npx",
      "args": ["-y", "@memsolus/mcp"],
      "env": {
        "MEMSOLUS_API_KEY": "mk_live_..."
      }
    }
  }
}

Zed

Add to Zed settings (Cmd+, > assistant > mcp):

{
  "context_servers": {
    "memsolus": {
      "command": {
        "path": "npx",
        "args": ["-y", "@memsolus/mcp"],
        "env": {
          "MEMSOLUS_API_KEY": "mk_live_..."
        }
      }
    }
  }
}

Any MCP-compatible client

The server runs over stdio by default. Point your client to:

npx -y @memsolus/mcp

With the environment variable MEMSOLUS_API_KEY set.

What your agent can do

15 tools total. All exposed automatically via MCP.

Use Cases

• Personalized assistants — Remember user preferences, past decisions, and context across sessions
• Multi-agent systems — Shared memory pools let agents collaborate with common context
• Knowledge management — Auto-extract entities and relationships from unstructured text
• Customer support — Recall full interaction history and customer preferences instantly
• Research agents — Accumulate findings across sessions, search by concept

Self-Hosting

If you're running your own Memsolus API instance, use the MEMSOLUS_API_URL variable to point to it:

{
  "mcpServers": {
    "memsolus": {
      "command": "npx",
      "args": ["-y", "@memsolus/mcp"],
      "env": {
        "MEMSOLUS_API_KEY": "mk_live_...",
        "MEMSOLUS_API_URL": "https://your-instance.example.com"
      }
    }
  }
}

Programmatic Usage

import { createServer } from '@memsolus/mcp';
import { StdioServerTransport } from '@modelcontextprotocol/sdk/server/stdio.js';

const server = createServer({
  baseUrl: 'https://api.memsolus.com',
  apiKey: process.env.MEMSOLUS_API_KEY,
});

const transport = new StdioServerTransport();
await server.connect(transport);

Configuration

Links

• Website
• Documentation
• API Reference
• Get API Key

License

MIT