memory-manager

Manage two-tier user memory with working and archived memory sections.

27|4|Updated Jun 12, 2025

Search Tags:#recall #memory #archive #memory-management #working-memory #fetch_memory_entries #recall_memory

Ready to run

npx skills add https://github.com/definableai/definable.ai --skill memory-manager-definableai

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: memory-manager
Source: https://github.com/definableai/definable.ai/tree/main/definable/definable/memory/v2/skill
Command: npx skills add https://github.com/definableai/definable.ai --skill memory-manager-definableai

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Two-tier memory management for AI agents: persistent working memory (always loaded) plus an indexed archived memory (on-demand) to recall, store, and forget information efficiently.

Core Features & Use Cases

Two-tier memory system: a always-loaded working memory for active facts and an archived memory for long-term context.
Structured recall and archiving: use recall_memory and fetch_memory_entries to retrieve relevant data and update memory sections.
Guardrails for context: enforce memory categories (user, project, reference, conversation) and update rules to prevent leakage or inconsistent state.

Quick Start

Create a working memory template for a user and configure recall/archival rules.

Frequently Asked Questions about memory-manager

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

FAQPage Schema

How do I maintain working memory and archived memory for an AI agent across sessions?▼

You can manage agent context across sessions by using a two-tier system that keeps working memory always loaded while indexing archived memory for on-demand recall.

What is the best way to structure memory categories to prevent context leakage in AI applications?▼

To prevent context leakage, enforce structured memory categories such as user, project, reference, and conversation, applying strict update and archiving rules to maintain state consistency.

How do I recall specific information from an indexed archive without loading the entire memory history?▼

You can recall specific information by using the fetch_memory_entries function to retrieve relevant data on demand from the indexed archived memory without loading the entire history.

When should I archive active facts from working memory into long-term storage?▼

You should archive active facts into long-term storage when they are no longer needed for immediate context but remain valuable for future recall, keeping working memory organized.

Does this two-tier memory management approach work for multi-project AI agents?▼

Yes, the two-tier memory management approach supports multi-project AI agents by categorizing context into project-specific sections and recalling relevant archived data on demand to maintain distinct boundaries.