Remembrances

Long-term memory for AI agents with privacy-first local embeddings and on-premise shared database (optional)

Remembrances MCP is a Model Context Protocol (MCP) server that gives your AI agents persistent, long-term memory. It uses local embeddings and SurrealDB to store and retrieve information securely, privacy-first, and without cloud dependencies.

About Remembrances

Learn more about the project, its philosophy, and the team behind it.

Read About

How it Works

Dive into the documentation to understand the architecture and integration.

Read Docs

Blog

Stay updated with the latest news, tutorials, and releases.

Read Blog

Privacy First

All embeddings generated locally with GGUF models. No data sent externally.

GPU Accelerated

Support for Metal (macOS), CUDA (NVIDIA), and ROCm (AMD) for blazing-fast performance.

Multiple Storage Layers

Key-value, vector/RAG, and graph database support with SurrealDB.

Quick Start

Get started with Remembrances MCP in minutes

For Linux or MacOSX

Execute the install script:



    curl -fsSL https://raw.githubusercontent.com/madeindigio/remembrances-mcp/main/scripts/install.sh | bash

For Windows

Use the linux version in WSL or with docker:



    docker run -it --rm \
    -v /path/to/data:/data \
    -v /path/to/kb:/knowledge-base \
    ghcr.io/madeindigio/remembrances-mcp:cpu