memory-manager

Persist AI session context across three memory tiers via CLI.

2|Updated Jan 29, 2026
One-click install
npx skills add https://github.com/hailingu/llm-prompts --skill memory-manager-hailingu
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: memory-manager
Source: https://github.com/hailingu/llm-prompts/tree/main/skills/memory-manager
Command: npx skills add https://github.com/hailingu/llm-prompts --skill memory-manager-hailingu

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) components.

What problem does it solve?

Memory continuity across interactions is hard without a centralized, versioned log; memory-manager provides a repo-local, three-tier memory system (L1 session logs, L2 theme notes, L3 global memory) with a simple CLI to persist, query, and promote knowledge.

Core Features & Use Cases

  • L1: Session logs capturing raw turns to memory/sessions
  • L2: Theme-based notes written to memory/<theme> with optional templates
  • L3: Global memory updates appended to memory/global.md for durable preferences
  • CSV memory: store, describe, and query tabular data with a manifest

Quick Start

Initialize a session with session-init and then persist a turn with persist-turn, optionally supplying extracted content and a theme.

Frequently Asked Questions about memory-manager

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

FAQPage Schema
How do I persist AI session context across multiple interactions?

To persist AI session context across multiple interactions, use a repo-local memory system that logs raw turns to session files and stores durable global preferences. This ensures conversational state remains intact between distinct agent runs.

What is a three-tier memory architecture for AI agents?

A three-tier memory architecture for AI agents separates context into L1 raw session logs, L2 extracted theme notes, and L3 durable global preferences. This structure allows interactive assistants to query and promote knowledge efficiently.

How do I save extracted notes and themes during an AI chat session?

To save extracted notes and themes during an AI chat session, use a CLI command that validates input themes and writes the extracted content to a dedicated memory directory, returning a structured JSON result for each operation.

Does this memory persistence approach work for long-running research projects?

Yes, repo-local memory persistence works for long-running research projects by maintaining a centralized, versioned log. It captures raw turns and appends durable global memory updates to keep continuity throughout the project lifecycle.

What is the best way to store and query tabular data in AI memory?

The best way to store and query tabular data in AI memory is using a CSV memory feature with a manifest. This allows you to store, describe, and query structured tabular data alongside standard session logs and global preferences.