loaditout.ai
SkillsPacksTrendingLeaderboardAPI DocsBlogSubmitRequestsCompareAgentsXPrivacyDisclaimer
{}loaditout.ai
Skills & MCPPacksBlog

memex

MCP Tool

xomidar/memex

Give your Claude Code the ability to actually remember you. Semantic memory, zero server deps.

Install

$ npx loaditout add xomidar/memex

Platform-specific configuration:

.claude/settings.json
{
  "mcpServers": {
    "memex": {
      "command": "npx",
      "args": [
        "-y",
        "memex"
      ]
    }
  }
}

Add the config above to .claude/settings.json under the mcpServers key.

About

Memex

Semantic memory for AI agents. An MCP server that gives your AI assistant a persistent, searchable knowledge base powered by vector embeddings.

Built for Claude Code. Works with any MCP-compatible client.

What is Memex?

In 1945, Vannevar Bush — MIT engineer and director of the US Office of Scientific Research and Development — published "As We May Think" in The Atlantic. He described a hypothetical device called the memex (memory + index):

> *"A memex is a device in which an individual stores all his books, records, and communications, and which is mechanized so that it may be consulted with exceeding speed and flexibility."*

The memex was never built. The technology didn't exist. But the concept — a personal device that extends human memory through associative retrieval — became the intellectual foundation for hypertext (Ted Nelson cited it), the World Wide Web (Tim Berners-Lee cited it), and personal computing.

This project brings Bush's vision to AI agents. Instead of losing context between sessions or guessing about user preferences, your AI assistant can store and retrieve knowledge semantically — by meaning, not keywords.

How It Works

Memex runs as a local MCP server that exposes four tools:

| Tool | Purpose | |------|---------| | ask | Semantic search — find relevant context by meaning | | memorize | Store new knowledge with auto-generated embeddings | | forget | Remove an entry by semantic match | | reflect | Audit what's stored about any topic |

Under the hood:

  • [LanceDB](https://lancedb.com/) — embedded vector database, no server needed
  • [OpenAI text-embedding-3-large](https://platform.openai.com/docs/guides/embeddings) — 3072-dimension embeddings, auto-generated on insert and search
  • MCP SDK — stdio transport, registers as a native tool in C

Tags

agent-memoryai-agentclaude-codeembeddingsknowledge-graphllmmcp-servermodel-context-protocolragsemantic-search

Reviews

Loading reviews...

Quality Signals

0
Installs
Last updated18 days ago
Security: AREADME

Safety

Risk Levelmedium
Data Access
read
Network Accessnone

Details

Sourcegithub-crawl
Last commit3/28/2026
View on GitHub→

Embed Badge

[![Loaditout](https://loaditout.ai/api/badge/xomidar/memex)](https://loaditout.ai/skills/xomidar/memex)