MCP Integration

The Model Context Protocol (MCP) lets AI assistants call external tools during a conversation. Neuroloom ships an MCP server that exposes memory operations as callable tools — so Claude and other MCP-compatible agents can store and retrieve memories without writing any glue code.

What is MCP?

MCP is an open standard for connecting AI models to external data sources and tools. When an AI assistant supports MCP, it can invoke tools like create_memory or search_memories as part of generating a response.

Learn more at modelcontextprotocol.io.

Prerequisites

Python 3.11+
A Neuroloom API key with write scope (Authentication)
An MCP-compatible client (Claude Desktop, Cursor, Windsurf, etc.)

Installation

pip install neuroloom-mcp

Available MCP Tools

Tool	Description
`create_memory`	Store a new memory with optional metadata
`search_memories`	Semantic search over stored memories
`get_memory`	Retrieve a specific memory by ID
`update_memory`	Update the content or metadata of a memory
`delete_memory`	Delete a memory permanently
`list_memories`	List memories with optional metadata filters

Configuring Claude Desktop

Open (or create) ~/Library/Application Support/Claude/claude_desktop_config.json and add the Neuroloom server:

{
  "mcpServers": {
    "neuroloom": {
      "command": "python",
      "args": ["-m", "neuroloom_mcp"],
      "env": {
        "NEUROLOOM_API_KEY": "nlk_live_your_key_here",
        "NEUROLOOM_WORKSPACE_ID": "ws_your_workspace_id"
      }
    }
  }
}

Restart Claude Desktop. You should see a hammer icon in the chat input confirming tools are available.

Configuring Other MCP Clients

Most MCP clients follow the same pattern — a command that starts the server process and env for credentials. Refer to your client's documentation for the exact config file location.

Using `uvx` (recommended for isolated environments)

{
  "mcpServers": {
    "neuroloom": {
      "command": "uvx",
      "args": ["neuroloom-mcp"],
      "env": {
        "NEUROLOOM_API_KEY": "nlk_live_your_key_here",
        "NEUROLOOM_WORKSPACE_ID": "ws_your_workspace_id"
      }
    }
  }
}

Example Conversation

Once connected, you can instruct Claude to use Neuroloom naturally:

"Remember that the deploy pipeline runs on Fridays at 22:00 UTC."

Claude calls create_memory with that content. Later:

"When does the deploy pipeline run?"

Claude calls search_memories with the query, retrieves the stored memory, and answers accurately.

Environment Variables

Variable	Required	Description
`NEUROLOOM_API_KEY`	Yes	Your API key
`NEUROLOOM_WORKSPACE_ID`	Yes	Target workspace
`NEUROLOOM_API_BASE_URL`	No	Override for self-hosted or staging

Troubleshooting

Tools not appearing in Claude Desktop Verify the config file path and JSON syntax. Restart Claude Desktop fully (not just the window).

401 Unauthorized errors Check that NEUROLOOM_API_KEY is set correctly and the key has write scope.

404 Not Found on workspace Confirm NEUROLOOM_WORKSPACE_ID matches a workspace your key can access.

Next Steps

Memories API — Understand the underlying REST API the MCP server calls
Authentication — Key scopes and workspace isolation

Ready to get started?

Start building with Neuroloom for free.

Start Building Free