engram

Organize AI memory across sessions with MEMORY.md, Knowledge Graph, and per-session notes.

1|Updated Feb 15, 2026
One-click install
npx skills add https://github.com/qwexs/engram --skill engram
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: engram
Source: https://github.com/qwexs/engram/tree/main
Command: npx skills add https://github.com/qwexs/engram --skill engram

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) and assets (resource) components.

What problem does it solve?

Solve the challenge of scalable, durable AI memory by providing a three-layer memory architecture with per-session notes, a Knowledge Graph, and curated long-term memory.

Core Features & Use Cases

  • Three-layer memory architecture (MEMORY.md, life/, memory/) for durable, organized knowledge
  • Knowledge Graph (life/) with flat three-folder structure (people/projects/archives) and per-entity summaries/items.json
  • QMD hybrid search (BM25 + vectors + rerank) for fast, relevant querying across memory layers
  • Heartbeat automation (extraction, synthesis, domain scans, maintenance) to keep knowledge fresh
  • Domain-based subagents with persistent memory and PROPOSAL workflow
  • Real-time inline extraction and conservative dedup and conflict checks
  • Memory decay and rotation mechanisms to surface durable knowledge over time

Quick Start

Install QMD, initialize Engram, and add a session to begin using Engram.

Frequently Asked Questions about engram

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I persist AI memory across sessions for long-term knowledge retention?

Persist AI memory across sessions by implementing a three-layer architecture combining MEMORY.md, a knowledge graph, and per-session notes to retain and organize durable knowledge.

What is the best way to organize a knowledge graph for multi-agent workflows?

Organize a knowledge graph for multi-agent workflows using domain-based subagents with persistent memory, flat three-folder structures, and per-entity summaries to enable scalable entity management.

How does memory decay work in AI knowledge graphs?

Memory decay works through rotation mechanisms and heartbeat-driven maintenance that automatically surface durable knowledge over time while deprecating stale, outdated information from the knowledge graph.

Can I use hybrid search to query persistent AI memory?

Yes, you can use QMD hybrid search to query persistent AI memory by combining BM25, vectors, and reranking to retrieve fast, relevant results across all active memory layers.

Do I need QMD to manage AI agent memory with heartbeat automation?

Yes, you need QMD installed to initialize the memory architecture and run heartbeat automation, which drives extraction, synthesis, and domain scans to keep knowledge fresh.