The Ultimate Memory Architecture — 21 neuroscience mechanisms orchestrating VividnessMem + VividEmbed
Project description
Mimir — The Ultimate Memory Architecture
21 neuroscience mechanisms orchestrating VividnessMem and VividEmbed into a unified memory system for AI agents and companions.
Named for the Norse god of wisdom who drank from the well of knowledge beneath Yggdrasil.
Key Features
| Mechanism | Citation |
|---|---|
| Flashbulb Memory | Brown & Kulik 1977 |
| Reconsolidation | Nader et al 2000 |
| State-Dependent Memory | Godden & Baddeley 1975 |
| Spreading Activation | Collins & Loftus 1975 |
| Retrieval-Induced Forgetting | Anderson 1994 |
| Zeigarnik Effect | Zeigarnik 1927 |
| Involuntary Recall | Berntsen 2009 |
| Temporal Gist | Reyna & Brainerd 1995 |
| Temporal / Episodic | Tulving 1972 |
| Visual / Mental Imagery | Kosslyn 1980 / Paivio 1986 |
| Huginn (Thought) | Pattern detection |
| Muninn (Memory) | Consolidation daemon |
| Yggdrasil (World Tree) | Memory graph |
| Völva's Vision | Dream synthesis |
| Novelty-Modulated Encoding | Ranganath & Rainer 2003 |
| Enhanced Drift Analysis | Cognitive bias detection |
| Hippocampal Pattern Separation | Computational neuroscience |
| Narrative Arc Tracking | Story structure |
| Task / Project Branch | Project memory |
| Hybrid Retrieval Bridge | BM25 + semantic + date + 5-signal re-rank |
| Encryption at Rest | Fernet + PBKDF2 |
Installation
# Core (zero required dependencies — everything degrades gracefully)
pip install vividmimir
# With the full Vivid engine stack
pip install vividmimir[all]
# Individual extras
pip install vividmimir[engines] # VividnessMem + VividEmbed
pip install vividmimir[visual] # Pillow for mental-imagery system
pip install vividmimir[encryption] # Fernet encryption at rest
Quick Start
from vividmimir import Mimir
m = Mimir(data_dir="my_mind")
# Store a memory
mem = m.remember("We talked about the northern lights in Iceland", "awe", 9)
# Retrieve context for an LLM
ctx = m.get_memory_context("Tell me about our travel plans")
print(ctx)
# Build a full context block (memories + emotions + lessons + tasks)
block = m.build_context_block("What were we discussing?")
Memory Types
- Episodic — event memories with emotion, vividness, and organic decay
- Social Profiles — per-person relationship tracking
- Procedural / Lessons — strategies learned from experience
- Volatile / Short-Term — facts that expire automatically
- Prospective / Reminders — time or context-triggered future actions
- Visual — mental imagery with Pillow integration
- Task / Project — branch-aware project memory
Part of the Vivid Ecosystem
| Package | Role |
|---|---|
| vividnessmem | Neurochemistry engine — organic decay, spaced repetition, emotion |
| vividembed | 389-d hybrid embeddings — semantic retrieval with emotion dimensions |
| vividmimir | Orchestrator — 21 mechanisms, hybrid retrieval, full memory architecture |
Documentation
Full API reference, benchmark results, and architecture details: github.com/Kronic90/Mimir
License
PolyForm Noncommercial 1.0.0 — free for personal and research use. Commercial use requires a separate license. Contact: GitHub Issues.
Project details
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
Filter files by name, interpreter, ABI, and platform.
If you're not sure about the file name format, learn more about wheel file names.
Copy a direct link to the current filters
File details
Details for the file vividmimir-1.0.1.tar.gz.
File metadata
- Download URL: vividmimir-1.0.1.tar.gz
- Upload date:
- Size: 54.8 kB
- Tags: Source
- Uploaded using Trusted Publishing? No
- Uploaded via: twine/6.2.0 CPython/3.12.7
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
a954d4ca3b167b1362bc887a5b5f4a60330578fbb05885e5b87883eed5ddb3a5
|
|
| MD5 |
96386011b7657f0dd204b739bdbd5793
|
|
| BLAKE2b-256 |
945eef3898118c705580f466137da4862b88184a0fc2110c954f7b36883a1a68
|
File details
Details for the file vividmimir-1.0.1-py3-none-any.whl.
File metadata
- Download URL: vividmimir-1.0.1-py3-none-any.whl
- Upload date:
- Size: 53.6 kB
- Tags: Python 3
- Uploaded using Trusted Publishing? No
- Uploaded via: twine/6.2.0 CPython/3.12.7
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
498b924352a70e972b533ddc37f27631e389ce34501038a86584948c692553ce
|
|
| MD5 |
c3583986ee468b0dca799f499c393d6a
|
|
| BLAKE2b-256 |
3d20cffa68e04359ae290e150077f4fa310abe84cfb38cc91646780c0f2e7c19
|