How does it manage memory relevance and decay?

It uses weighted memory nodes, where importance increases with use. A time-based decay system automatically prunes stale memories, ensuring that only relevant knowledge persists in your graph.

Can I use it with my existing Claude conversations?

Absolutely. The tool supports chat history ingestion, allowing you to review your past Claude Code sessions and extract valuable knowledge to enrich your memory graph.

What is Adaptive Memory Graph?

Adaptive Memory Graph is an MCP plugin designed to give Claude persistent, intelligent memory across sessions. It stores knowledge as an evolving, weighted graph of interconnected nodes.

Is my stored knowledge secure?

Yes, all knowledge is protected with AES-256-GCM encryption. For macOS users, the encryption key is securely stored and managed via the macOS Keychain.

Adaptive Memory Graph

Name: Adaptive Memory Graph
Author: raskolnikovdd

0•

Provides Claude with persistent, intelligent memory across sessions, storing knowledge as an evolving, weighted graph.

The Adaptive Memory Graph is an MCP server plugin designed to equip Claude with persistent and intelligent memory capabilities across multiple conversational sessions. It functions by storing knowledge as weighted, interconnected nodes in a dynamic graph that evolves with every interaction. As Claude utilizes specific nodes, their weight increases, while unused memories gradually decay and are eventually archived, ensuring that only relevant information persists. This system features encrypted storage for security, organizes nodes by domain, logs session activity, and allows for the ingestion of past chat histories to enrich Claude's knowledge base, ultimately enabling a more knowledgeable and contextually aware AI assistant.

主な機能

01Weighted memory nodes for importance and relevance

02Time-based decay for self-pruning of stale memories

03AES-256-GCM encrypted storage with macOS Keychain support

04Cross-domain connections to link related knowledge

05Chat history ingestion to review and extract knowledge from past sessions

060 GitHub stars

ユースケース

01Maintaining persistent context and knowledge across diverse AI interactions

02Organizing and extracting insights from past conversational data to enhance AI knowledge

03Equipping AI assistants like Claude with long-term, evolving memory capabilities