Memory operating system for LLMs and AI agents that stores, retrieves, and manages long-term memory as an editable graph.
12.3K+103 this week1.1KTypeScriptApache-2.0Updated 3 days ago
- Stars
- 623
- Forks
- 81
- Open issues
- 65
- Last commit
- 3 days ago
What is it?
Memoria is a persistent memory layer for AI agents that tracks every memory change so it can be audited, branched, and reverted. It stores facts, preferences, decisions, and session summaries so an agent can reuse them across conversations, and it exposes these operations as MCP tools or a REST API.
The version control model works like Git for memory: you can take snapshots, create isolated branches to test changes, merge them back after review, or roll back to an earlier state. These features run on the Copy-on-Write engine of MatrixOne, which the project uses for storage and combined vector and full-text retrieval.
Memoria can run as a hosted cloud service with a token, or self-hosted with Docker and a local embedding option so data stays on your machine. It works with MCP-compatible tools such as Cursor, Claude Code, Codex, Gemini CLI, and Kiro, and includes plugins for OpenClaw and Hermes.
Key features
- 01Snapshots, branches, merges and point-in-time rollback for memory data
- 02Vector and full-text hybrid retrieval for finding memories by meaning
- 03Automatic contradiction detection and quarantine of low-confidence memories
- 04Audit trail with a snapshot and provenance for every memory change
- 05Cloud or self-hosted deployment, with an optional local embedding model
- 06Works with any MCP-compatible agent through the memoria CLI
Use cases
- Giving a coding assistant persistent knowledge of project decisions and user preferences across sessions
- Testing a risky change to an agent's memory on a branch before merging it into main
- Rolling back an agent's memory after bad or incorrect information was stored
- Keeping agent memory fully on-premises for teams with strict data control requirements
Quick start
- 1
Sign up at thememoria.ai and get your token
- 2
Run the official install script from the README
- 3
Run memoria init, select Remote mode and paste your token
memoria init -i
Things to know
- The recommended quick start depends on the hosted Memoria Cloud service and a sign-up token.
- Self-hosting requires Docker to run MatrixOne and the Memoria API.
In collections
Topics
FAQ
- What is memoria?
- Persistent memory layer for AI agents with Git-style snapshots, branches, merges and rollback, backed by MatrixOne. Memoria is a persistent memory layer for AI agents that tracks every memory change so it can be audited, branched, and reverted. It stores facts, preferences, decisions, and session summaries so an agent can reuse them across conversations, and it exposes these operations as MCP tools or a REST API.
- Is memoria free to use? What license does it have?
- memoria is open source under the Apache-2.0 license. Check the license terms before using it in commercial products.
- Is memoria still maintained?
- Actively maintained. The last commit to memoria was on Nov 4, 2026.
- What can I use memoria for?
- Giving a coding assistant persistent knowledge of project decisions and user preferences across sessions Testing a risky change to an agent's memory on a branch before merging it into main Rolling back an agent's memory after bad or incorrect information was stored Keeping agent memory fully on-premises for teams with strict data control requirements
- How do I install memoria?
- Sign up at thememoria.ai and get your token Run the official install script from the README Run memoria init, select Remote mode and paste your token: memoria init -i
- Where can I get memoria?
- The source code and installation instructions are on GitHub: https://github.com/matrixorigin/memoria
Summary generated by AI from the project's README. GitHub stats refresh daily.
Related projects
A team-level memory hub for AI Agents that turns chats, docs, and code into shared, governed memory assets usable across agent frameworks.
30.6K+315 this week3KTypeScriptOtherUpdated 3 days ago
A Rust framework for building composable AI agent runtimes that combine models, tools, memory and other agents.
448+5 this week55RustApache-2.0Well maintainedUpdated 18 days ago
Memory infrastructure for stateful AI agents that reasons over conversations to model users, agents and groups over time.
8K+144 this week973PythonAGPL-3.0Updated 16 hours ago
Local-first AI memory system that stores conversation history verbatim and retrieves it with semantic search, exposed as an MCP server and CLI.
60.4K+287 this week7.7KPythonMITUpdated 17 hours ago
Local, evidence-first memory runtime for AI agents that filters writes, tracks fact changes over time, and explains each recalled result.
69257PythonApache-2.0Updated 3 months ago