Skip to main content

​
What is MCP

MCP (Model Context Protocol) is a protocol for directly integrating Neuradex memory capabilities into LLMs (Claude, GPT, etc.).
With MCP, AI assistants can directly reference and update the Neuradex knowledge base.

​
Features

Seamless Integration

Access Neuradex directly from Claude Desktop or VS Code

Real-time Search

Search and reference knowledge during conversations

Context Optimization

Automatically assemble optimal context within token budget

Knowledge Management

Full CRUD operations: create, update, and delete

Remote Connection

Connect to cloud-hosted MCP server via HTTP mode

OAuth Authentication

Secure token-based authentication for remote access

​
Connection Modes

The Neuradex MCP server supports two transport modes. See the setup guide for detailed configuration instructions.
MCP clients (like Claude Desktop) launch the server locally and communicate via standard I/O.
  • Easy to set up
  • Authenticated with access key
  • Ideal for personal development

​
Use Cases

​
Long-term Memory for AI Assistants

User: Tell me about the return policy we discussed last week

Claude: [Searches Neuradex]
        I checked last week's conversation records and related knowledge.
        Returns are accepted within 30 days of product arrival...

​
Access to Internal Knowledge

User: Check the specifications for the new authentication system

Claude: [Searches Neuradex]
        I found the authentication system specification document.
        It uses OAuth 2.0 with JWT tokens for authorization...

​
Learning from Conversations

User: Let's share an agenda before all future meetings. Remember this.

Claude: [Saves to Neuradex]
        Understood. I've saved the rule "Share agenda before meetings"
        as knowledge.

​
Available Tools

The MCP server provides 16 tools (see the tool reference for details):
CategoryToolDescription
Projectlist_projectsList accessible projects
Sessionswitch_projectSwitch active project
Sessioncurrent_projectShow current active project
Knowledgecreate_knowledgeCreate new knowledge
Knowledgesearch_knowledgeSemantic search
Agentic Searchsearch_agenticCross-layer search (single array)
Knowledgeview_knowledgeGet knowledge details
Knowledgelist_knowledgeList all knowledge
Knowledgeupdate_knowledgeUpdate knowledge
Knowledgedelete_knowledgeDelete knowledge
Episodeslist_episodesSearch Q&A history
Contextget_contextGet optimized context
Contextexplore_relatedExplore related knowledge
Librarianlist_librarian_sessionsList Librarian sessions
Librarianget_librarian_sessionGet Librarian session details

​
Architecture

​
Next Step

Setup

Install and configure the MCP server