Basic Memory

Basic Memory

Local-first MCP server enabling LLMs to read/write a local Markdown knowledge graph.

2,113
Stars
131
Forks
20
Releases

Overview

Basic Memory is an MCP-enabled local-first knowledge management solution that lets AI assistants read and write to your personal knowledge base stored as Markdown files on your computer. By implementing the Model Context Protocol (MCP), the server exposes structured access to a semantic graph of Entities, Observations, and Relations that can be traversed by compatible LLMs. Notes are created and updated in real time in Markdown, with a local SQLite database indexing and searching the knowledge graph, ensuring persistent memory across conversations. The system supports bidirectional synchronization between the file system and the knowledge graph, so humans edit files and LLMs enrich the graph, while keeping everything under your control and privacy-preserving. It uses memory:// URLs to reference entities across tools and conversations. The server provides tools for content management, knowledge graph navigation, and search/discovery via MCP, and can be used within environments like VS Code or Claude Desktop through configured MCP endpoints. Cloud-based features are available via subscription, including authentication, bidirectional cloud sync, integrity checks, and mounting cloud storage, while local data remains stored as Markdown files by default.

Details

Owner
basicmachines-co
Language
Python
License
GNU Affero General Public License v3.0
Updated
2025-12-07

Features

Local Markdown knowledge base with semantic graph

Stores notes as Markdown and structures them into Entities with Observations and Relations to form a traversable knowledge graph.

Bidirectional synchronization

Keeps local Markdown files and the knowledge graph in sync so changes propagate in both directions.

MCP exposure and tooling

Implements the Model Context Protocol (MCP) to enable AI integration, exposing traversal and manipulation tools.

Memory URLs across tools

Uses memory:// URLs to reference entities across different tools and conversations.

SQLite-backed indexing and search

Indexes and enables fast search over the knowledge graph using a local SQLite database.

Real-time persistence

Notes are saved locally in real time, enabling immediate consistency across sessions.

Multiple projects support

Supports multiple memory projects with project-level configuration and management commands.

Cloud integration (optional)

Cloud login, bidirectional sync, integrity checks, and cloud mounting available via subscription, with local Markdown data preserved by default.

Audience

Solo knowledge workersMaintain a persistent local knowledge base and memory across conversations with LLMs using MCP.
Developers/IntegratorsIntegrate MCP-based knowledge graph traversal/manipulation into workflows (e.g., via VS Code or Claude Desktop).
Productivity enthusiastsOrganize notes semantically, link related topics, and perform efficient searches with SQLite.
Cloud-enabled usersExtend capabilities to cross-device syncing and cloud storage through a subscription.

Tags

local-firstmarkdownknowledge-graphMCPModel Context ProtocolSQLitesemantic-graphobservationsrelationsentitiesmemoryObsidianClaude Desktopcloud-sync