---

What Is MNEMOSYNTH?

LLMs forget everything after each session. MNEMOSYNTH gives AI a second brain — it stores, verifies, and retrieves knowledge with cognitive memory patterns inspired by neuroscience.

pip install mnemosynth     # One command — everything works
mnemosynth serve           # Start MCP server for Claude/Cursor/Windsurf

Zero-config. No Docker. No external databases. No API keys. Everything runs locally under ~/.mnemosynth/.

---

🏗️ Project Structure

mnemosynth/
├── mnemosynth-core/        🐍 Python package (PyPI)
├── mnemosynth-extension/   🧩 VS Code extension (Marketplace)
└── mnemosynth-web/         🌐 Landing page & docs (React/Vite)

Project	Stack	Live At
mnemosynth-core	Python · LanceDB · NetworkX · MCP	PyPI
mnemosynth-extension	TypeScript · VS Code API · Webview	Marketplace
mnemosynth-web	React · Vite · Framer Motion · TypeScript	Deploy to Netlify/Vercel

---

✨ Core Features

Three-Tier Cognitive Memory

Tier	Brain Analog	What It Stores
Episodic 🔵	Hippocampus	Events, conversations, timestamped history
Semantic 🟢	Neocortex	Verified facts, entity relationships
Procedural 🟠	Cerebellum	Tools, schemas, workflows

Anti-Hallucination Engine

Feature	Description
🧊 Ebbinghaus Decay	Stale memories lose confidence over time
⚔️ Contradiction Detection	DeBERTa NLI flags conflicting facts
💜 Sentiment Scoring	Emotional memories rank higher in retrieval
🛡️ Immune System	Blocks prompt injections, rate limits, quarantines
📊 Corroboration	Repeated observations boost confidence
💤 Dream Consolidation	HDBSCAN clusters → promote to verified facts

Agentic Framework Adapters

from mnemosynth.adapters.crewai import get_crewai_tools       # CrewAI
from mnemosynth.adapters.langchain import MnemosynthMemory     # LangChain
from mnemosynth.adapters.autogen import get_autogen_tools      # AutoGen
from mnemosynth.adapters.memory_bus import MemoryBus           # Cross-Agent Bus

Causal Memory Chains

from mnemosynth.engine.causal import CausalChainEngine
engine = CausalChainEngine(brain)
engine.record_chain(
    decision="Use SQLite", reasons=["Zero-config"], outcome="Users can pip install"
)

Agent-to-Agent (A2A) Protocol

from mnemosynth.a2a import A2AProtocol
proto = A2AProtocol(brain)
proto.share_memory("User prefers Python", source_agent="researcher", visibility="shared")

OpenTelemetry Tracing

from mnemosynth.telemetry import instrument_brain
instrument_brain(brain)  # All operations now emit traces + metrics

Production DB Backends

from mnemosynth.stores.qdrant_store import QdrantEpisodicStore     # Qdrant
from mnemosynth.stores.falkordb_store import FalkorDBSemanticStore  # FalkorDB
from mnemosynth.stores.postgres_store import PostgresStore          # PostgreSQL
---

## 🚀 Upcoming Features

- **Native IDE Integrations** - Deep integration with popular IDEs beyond VS Code
- **Chronological Timeline & Conflict Resolution** - Enhanced timeline view with advanced conflict detection and resolution
- **Privacy First** - End-to-end encryption and privacy-focused architecture

---

## 🚀 Quick Start

### Python API
```python
from mnemosynth import Mnemosynth
brain = Mnemosynth()
brain.remember("User prefers Python and dark mode")
results = brain.recall("What does the user prefer?")
brain.dream()

MCP Server

{ "mcpServers": { "mnemosynth": { "command": "mnemosynth", "args": ["serve"] } } }

VS Code Extension

code --install-extension Vasudev-Jaiswal.mnemosynth

---

🛠️ CLI

mnemosynth serve              # Start MCP server
mnemosynth stats              # Memory dashboard
mnemosynth search "query"     # Semantic search
mnemosynth inspect            # Browse memory tree
mnemosynth dream              # Run consolidation
mnemosynth export -o out.json # Export to JSON

---

📁 Development

# Core
cd mnemosynth-core && pip install -e ".[dev]" && pytest

# Extension
cd mnemosynth-extension && npm install && npm run compile

# Web
cd mnemosynth-web && npm install && npm run dev

---

🔗 Links

Resource	URL
PyPI Package	pypi.org/project/mnemosynth
VS Code Extension	VS Code Marketplace
GitHub Repository	github.com/vasudevjaiswal/mnemosynth
Web Documentation	mnemosynth.tech

---

📄 License

Apache 2.0 — © 2025 Vasudev Jaiswal

---