skills.rest
nicolehaugen
nicolehaugenCOMMUNITY
Explore all skills by this author →
GitHub

tiered-memory

Implement a three-tier memory model with hot, cold, and wiki tiers.

Updated Apr 8, 2026
Search Tags:#performance optimization#memory management#agent memory#context reduction#Squad agent
One-click install
npx skills add https://github.com/nicolehaugen/CustomMetricsDashboard --skill tiered-memory-nicolehaugen
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: tiered-memory
Source: https://github.com/nicolehaugen/CustomMetricsDashboard/tree/main/.squad/templates/skills/tiered-memory
Command: npx skills add https://github.com/nicolehaugen/CustomMetricsDashboard --skill tiered-memory-nicolehaugen

SYSTEM DOCUMENTATION & REQUIREMENTS

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

What problem does it solve?

This Skill reduces the size of agent context payloads by eliminating redundant information, resulting in improved performance and reduced memory usage.

Core Features & Use Cases

  • Hot Tier: Current session context (2-4KB), always loaded.
  • Cold Tier: Summarized cross-session history (8-12KB), loaded on demand.
  • Wiki Tier: Durable structured knowledge, loaded on demand.
  • Use Case: Implementing a three-tier memory model in a Squad agent to optimize context management and enhance performance.

Quick Start

To use the tiered-memory skill, spawn an agent with the following template:

## Memory Context

### Hot (current session)
{hot_context}

Frequently Asked Questions about tiered-memory

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

FAQPage Schema
How do I reduce agent context bloat and improve memory management performance?▼

Agent context bloat can be reduced by 20-55% using a three-tier memory model that separates data into hot, cold, and wiki tiers, eliminating redundant information from payloads to optimize performance.

What is the best way to structure agent memory for different context types?▼

Structuring agent memory involves a three-tier model: a hot tier for current session context, a cold tier for summarized cross-session history, and a wiki tier for durable structured knowledge, each loaded on demand or always active.

Does the tiered memory model require a Scribe agent to function?▼

Yes, implementing this three-tier memory model requires integration with a Scribe agent to handle tier promotion, moving summarized information between the hot, cold, and wiki context tiers effectively.

How do I implement a three-tier memory model in a Squad agent?▼

To implement tiered memory in a Squad agent, spawn an agent using a memory context template that defines the hot tier for the current session, then integrate the Scribe agent to manage cold and wiki tier promotion.

What are the size limits for hot and cold memory context tiers?▼

The hot tier holds current session context at 2-4KB and is always loaded, while the cold tier stores summarized cross-session history at 8-12KB and is loaded on demand to manage payload size.

Why does my agent experience high memory usage during long sessions?▼

High memory usage often results from unoptimized context payloads containing redundant information; applying a tiered memory model reduces this bloat by up to 55% by managing active versus on-demand context.

Related Skills

Explore More Software Engineering Skills→
acmelemeacmeleme

tiered-memory

Implement a three-tier hot, cold, and wiki memory model for Squad agents.

Official
Advanced
CommunityToolkitCommunityToolkit
619

tiered-memory

Implement three-tier Hot, Cold, and Wiki memory for Squad agents.

Official
Advanced
m16khbm16khb
1

draft-wiki-promoter

Review and promote approved draft-wiki candidates into a repo-local exported directory.

Community
Intermediate
teslasoft-deteslasoft-de
1

context-budget

Manage context budgets with progressive disclosure and four-tier loading.

Community
Advanced