memory-mcp

The Problem

Every new chat starts from scratch. You explain your architecture again. You paste the same patterns again. Your context window bloats with repetition.

Other memory solutions help, but they still require tool calls for every lookup — adding latency and eating into Claude's thinking budget.

Memory MCP fixes this with a two-tier architecture:

1. Hot cache (0ms) — Frequently-used knowledge auto-injected into context before Claude even starts thinking. No tool call needed.
2. Cold storage (~50ms) — Everything else, searchable by meaning via semantic similarity.

The system learns what you use and promotes it automatically. Your most valuable knowledge becomes instantly available. No manual curation required.

Before & After

Install

# Install package
uv tool install hot-memory-mcp   # or: pip install hot-memory-mcp

# Add plugin (recommended)
claude plugins add michael-denyer/memory-mcp

The plugin gives you auto-configured hooks, slash commands, and the Memory Analyst agent. MLX is auto-detected on Apple Silicon.

{
  "mcpServers": {
    "memory": {
      "command": "memory-mcp"
    }
  }
}

Restart Claude Code. The hot cache auto-populates from your project docs.

First run: Embedding model (~90MB) downloads automatically. Takes 30-60 seconds once.

How It Works

flowchart LR
    subgraph LLM["Claude"]
        REQ((Request))
    end

    subgraph Hot["HOT CACHE · 0ms"]
        HC[Session context]
        PM[(Promoted memories)]
    end

    subgraph Cold["COLD STORAGE · ~50ms"]
        VS[(Vector search)]
        KG[(Knowledge graph)]
    end

    REQ -->|"auto-injected"| HC
    HC -.->|"draws from"| PM
    REQ -->|"recall()"| VS
    VS <-->|"related"| KG

The hot cache (~10 items) is injected into every request — it combines recent recalls, predicted next memories, and top promoted items. Promoted memories (~20 items) is the backing store of frequently-used memories. Memories used 3+ times auto-promote; unused ones demote after 14 days.

What Makes It Different

Most memory systems make you pay a tool-call tax on every lookup. Memory MCP's hot cache bypasses this entirely — your most-used knowledge is already in context when Claude starts thinking.

The Engram Insight: Human memory doesn't search — frequently-used patterns are already there. That's what hot cache does for Claude.

Quick Reference