agent-memory

Save and organize user memories with frontmatter summaries across conversations.

Updated Aug 27, 2026
One-click install
npx skills add https://github.com/mgr-kb/mgr-dotfiles --skill agent-memory-mgr-kb
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: agent-memory
Source: https://github.com/mgr-kb/mgr-dotfiles/tree/main/claude/skills/agent-memory
Command: npx skills add https://github.com/mgr-kb/mgr-dotfiles --skill agent-memory-mgr-kb

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This skill provides a persistent memory space that survives across conversations, enabling users to save, recall, and organize important information for ongoing work.

Core Features & Use Cases

  • Proactive memory capture: Save research findings, patterns, architectural decisions, and other valuable knowledge.
  • Cross-session recall: Access and refresh context before starting related work.
  • Organization & lifecycle: Create category-based memories, update status, consolidate related items, and clean up outdated information.

Quick Start

Save a memory about a finding by creating a new memory entry under .claude/skills/agent-memory/memories with a short summary frontmatter and the full content.

Frequently Asked Questions about agent-memory

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

FAQPage Schema
How do I persist context and notes across different chat sessions?

To persist context across sessions, you need a memory store that saves information with summary metadata. This skill captures research findings and architectural decisions, organizing them by category to support recall when starting related work later.

What is the best way to organize long-term knowledge management for ongoing projects?

The best way to organize long-term knowledge management is using category-based grouping with frontmatter summaries. This approach supports search, tagging, and lifecycle management, allowing you to consolidate related items and clean up outdated information.

How do I proactively capture and recall memories for ongoing work?

You capture and recall memories by saving entries into a dedicated memories directory. Each entry requires frontmatter with summary and created fields, enabling proactive capture of patterns and automatic context refresh when resuming related tasks.

Can I use category-based organization to manage my knowledge lifecycle?

Yes, category-based organization supports the full knowledge lifecycle. You can create categorized memories, update their status over time, consolidate related items together, and clean up outdated information to maintain an organized persistence store.

Does cross-session memory persistence require a specific file structure?

Cross-session memory persistence requires a specific directory structure under .claude/skills/agent-memory/memories. Each memory entry must include frontmatter with summary and created fields to support flexible organization and search.

Why does my agent lose context when starting a new conversation session?

Agents lose context because standard conversations do not retain information. You need a persistent memory space that survives across sessions, allowing you to save valuable knowledge and recall it at the start of new related work.