Model Context Protocol

Give Claude Persistent Memory

Knol's MCP server lets Claude Desktop, Cursor, Windsurf, and any MCP-compatible tool remember users, learn preferences, and build knowledge across sessions.

Get Started FreeRead the Docs

Install in One Command

npx @aiknol/knol-mcp-server

No Docker, no PostgreSQL setup required for local use. Just run and connect.

How It Works

1

Install the MCP Server

Run the npx command or add Knol to your Claude Desktop MCP config. The server starts automatically and connects to your local Knol instance.

2

Claude Learns From Conversations

As you chat, Knol automatically extracts facts, preferences, and relationships. These are stored as structured memories with temporal context.

3

Every Session Gets Smarter

Next time you open Claude, it already knows your projects, preferences, and context. No more repeating yourself. Memory persists across sessions.

Claude Desktop Configuration

Add this to your Claude Desktop MCP configuration file:

{
  "mcpServers": {
    "knol-memory": {
      "command": "npx",
      "args": ["@aiknol/knol-mcp-server"],
      "env": {
        "KNOL_API_URL": "http://localhost:3000",
        "KNOL_API_KEY": "your-api-key"
      }
    }
  }
}

Config file location: ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%/Claude/claude_desktop_config.json (Windows)

Available MCP Tools

Once connected, Claude can use these tools to manage persistent memory:

knol_store_memory

Store a new memory from the current conversation — facts, preferences, decisions, or context.

knol_search_memory

Search past memories using natural language. Uses hybrid retrieval (vector + BM25 + graph) for accurate results.

knol_get_user_context

Retrieve a summary of everything known about the current user — preferences, history, and relationships.

knol_list_memories

List recent memories with optional filters by type, date range, or topic.

knol_delete_memory

Remove a specific memory by ID. Supports GDPR right-to-erasure compliance.

knol_graph_query

Traverse the knowledge graph to find relationships between entities (people, projects, concepts).

What You Can Build

Personal AI Assistant

Claude remembers your projects, coding preferences, writing style, and ongoing tasks. Ask "what was I working on last week?" and get accurate answers.

Team Knowledge Base

Share memory across team members. Decisions, architecture choices, and project context are captured and retrievable by anyone on the team.

Cursor / Windsurf Integration

Your AI coding assistant remembers your codebase patterns, tech stack preferences, and past debugging sessions. Context carries across projects.

Customer Support Agent

Build support bots that remember every customer interaction. Prior tickets, preferences, and account details are available instantly.

Why Knol for MCP

Most MCP memory servers store flat key-value pairs or simple text blobs. Knol is a full context engineering platform with hybrid retrieval, knowledge graphs, memory decay, and conflict detection — all exposed through MCP tools.

<5ms
Memory retrieval latency
Hybrid
Vector + BM25 + Graph search
Temporal
Facts evolve over time

Ready to Give Your AI Persistent Memory?

Knol is open-source and self-hostable. Deploy on your infrastructure, keep your data private, and give every AI tool you use persistent memory.

Star on GitHubTry the DemoView Pricing