Skip to main content
MCP server CX Codex

SuperLocalMemory — Claude Code MCP server: setup guide

A local-first memory layer for AI agents that keeps project context on your machine. It checks facts and admits when information is missing instead of guessing.

Install into Claude Code · Platform CLI All methods and platforms ↓
claude mcp add superlocalmemory -- npx -y superlocalmemory
Open on GitHub

What SuperLocalMemory does

Agents lose their context once a session ends. Standard vector stores often pull irrelevant text snippets. SuperLocalMemory maintains a local store on your machine. In Mode A, it works without cloud API calls, Docker, or remote accounts. Text queries use a local nomic model.

You can also index screenshots, images, and PDF documents. This feature uses Google's EmbeddingGemma 2 model. It unloads after 30 minutes of inactivity. Running this media mode requires at least 16 GB of system RAM and roughly 1.5 GB of disk space.

The server connects to coding tools through the slm CLI or a standard stdio command. When an answer is missing, your assistant will not bluff.

What it helps with

  • Retain architectural decisions across different agent sessions
  • Search local design screenshots and PDF specifications
  • Stop coding models from hallucinating missing project rules
  • Maintain persistent agent state without sending data to external APIs

Example prompts for the agent

  • Remember that we use PostgreSQL 16 for all new database tables
  • Find the authentication flowchart diagram I added last Tuesday
  • Check our memory for decisions regarding API rate limits

How to install SuperLocalMemory

Platform CLI

claude mcp add superlocalmemory -- npx -y superlocalmemory

Step by step, with commands for each platform: How to add an MCP server to Claude Code, Codex, Cursor, VS Code, Gemini CLI and more

Questions about SuperLocalMemory

How do I install SuperLocalMemory in Claude Code?

Run claude mcp add superlocalmemory -- npx -y superlocalmemory in your terminal.

What does SuperLocalMemory do?

It provides local long-term memory for coding agents across sessions. It recalls decisions, diagrams, and files while refusing to invent answers when records do not exist.

What is required to search media files?

Processing images and PDF documents requires around 1.5 GB of storage space and a machine with at least 16 GB of RAM.

Similar MCP servers

All MCP servers for Agent setup: memory, context, orchestration →

Project sponsors

Companies that keep this analytics open
Gold sponsor

Impressions, clicks and code copies — open statistics. We keep the numbers public for the same reason NeBlask itself is public. Become a sponsor