Skip to content

Welcome to MIRIX

MIRIX is a memory system for agents. It captures, structures, and retrieves memories so your agents can stay consistent over time.

Important Update: 0.1.6 (Main) vs 0.1.3 (Desktop Agent)
Starting with 0.1.6, the main branch is a brand-new release line where Mirix is a pure memory system that can be plugged into any existing agents. The desktop personal assistant (frontend + backend) has been deprecated and is no longer shipped on main. If you need the earlier desktop application with the built-in agent, use the desktop-agent branch.

  • Get Started

    Overview and a fast path to your first memory write.

    Overview

  • Memory Write

    Configure LLMs, embeddings, and retention policies.

    Configuration

  • Memory Search

    Query memories with keyword and embedding search.

    Search

  • Contributing

    Learn how to contribute to the MIRIX project and join our community.

    Contribute

How MIRIX Works

flowchart TD
    A[Inputs] --> B[Meta Agent]
    B --> C{Content Analysis}
    C --> D[Core Memory<br/>Personal Info]
    C --> E[Episodic Memory<br/>Activities]
    C --> F[Semantic Memory<br/>Knowledge]
    C --> G[Procedural Memory<br/>Workflows]
    C --> H[Resource Memory<br/>Documents]
    C --> I[Knowledge Vault<br/>Credentials]

    J[Agent Query] --> K[Retrieval]
    K --> L[Memory Search]
    D --> L
    E --> L
    F --> L
    G --> L
    H --> L
    I --> L
    L --> M[Intelligent Response]

Use Cases

Agent Memory

Persist key facts and decisions so agents remain consistent across sessions.

Retrieval and Recall

Query episodic and semantic memory with keyword or embedding search.

Structured Knowledge

Store procedures, resources, and core facts in dedicated memory types.

Multi-Agent Systems

Plug MIRIX into different agents without rebuilding memory pipelines.

System Requirements

  • Python 3.11 or later
  • PostgreSQL 17 (recommended) or SQLite
  • An API key for your LLM provider

Ready to add memory to your agents?

Get Started Now

Key Capabilities

Memory System

  • Six memory components with dedicated agents
  • Configurable retention and decay policies
  • Structured writes from conversation input

Multi-Agent Architecture

  • 8 specialized agents working collaboratively
  • 6 memory components for organized data storage
  • Coordinated workflow for efficient processing
  • PostgreSQL-native BM25 search
  • Vector similarity search using embeddings
  • Field-specific search across all memory types

Privacy & Security

  • All long-term data stored locally
  • User-controlled privacy settings
  • Enterprise-grade PostgreSQL security