agent-memory-management

Implement stateful AI agents with sessions and tiered memory storage.

Updated Apr 13, 2026
One-click install
npx skills add https://github.com/bermudi/skills --skill agent-memory-management-bermudi
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: agent-memory-management
Source: https://github.com/bermudi/skills/tree/main/agent-memory-management
Command: npx skills add https://github.com/bermudi/skills --skill agent-memory-management-bermudi

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

Stateful agents require durable memory across sessions to preserve user preferences, context, and coordination between subagents, enabling personalized and reliable long-running interactions.

Core Features & Use Cases

  • Memory filesystem design with a three-tier layout (System, Progressive, Recall) to balance recall and cost
  • Session management and cross-agent memory coordination with versioned storage and provenance tracking
  • Retrieval and governance features including guardrails, anti-poisoning, and provenance to ensure trustworthiness

Quick Start

Initialize a memory suite by creating a system/ folder with identity files and a MEMORY.md for long-term facts and load it into your agent.

Frequently Asked Questions about agent-memory-management

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

FAQPage Schema
How do I build stateful AI agents that remember context across sessions?

Build stateful agents by using Sessions for short-term history and Memory for long-term persistence across multi-turn interactions. This preserves user preferences, decisions, and contextual facts, enabling personalized and reliable long-running interactions.

How does multi-agent memory coordination work with versioned storage?

Multi-agent memory coordination works by applying versioned storage with provenance tracking to shared memory. This ensures that subagents can preserve context and coordination reliably while maintaining a trustworthy record of memory changes.

What is the best way to structure a memory filesystem for AI agents?

The best way to structure a memory filesystem is using a three-tier layout: System, Progressive, and Recall. This design balances recall effectiveness and cost by separating identity files, long-term facts, and extracted contextual data.

How do I prevent memory poisoning in multi-agent systems?

Prevent memory poisoning by applying retrieval guardrails and provenance tracking within the extraction-consolidation-retrieval pipeline. These governance features ensure trustworthiness and stop corrupted or malicious data from persisting.

Do I need a specific framework to implement long-term agent memory?

You do not need a specific framework to implement long-term agent memory. You can initialize a memory suite by creating a system/ folder with identity files and a MEMORY.md for long-term facts, then load it into your agent.

When should I use a three-tier memory layout instead of standard session history?

Use a three-tier memory layout instead of standard session history when your agents need durable long-term persistence alongside short-term recall. It is essential for multi-turn interactions where remembering user preferences and contextual facts across sessions is required.