tiered-memory

Community

Keep knowledge across resets with tiered memory.

AuthorCheggin
Version1.0.0
Installs0

System Documentation

What problem does it solve?

Three-tier memory system stores context on disk in .harness/memory/{hot,warm,cold} to prevent context pollution and survive resets.

Core Features & Use Cases

  • Hot tier: always loaded (max 2K tokens) and injected into every agent session; contains current task description, active feature checklist, and blockers; auto-summarizes when cap is reached.
  • Warm tier: seven-day TTL; loaded on demand; stores recent architectural decisions, recent GitHub issues, and common error patterns; entries expire after 7 days.
  • Cold tier: permanent; source of completed feature summaries, historical decisions, and post-mortems; surfaced when needed.
  • Promotion/demotion: moves items between tiers as relevance changes (hot-to-warm/cold; cold-to-hot when reactivated).
  • Persistence: integrates with context-reset-handler; PostStart loads hot memory and provides access to warm and cold retrieval.

Quick Start

Initialize and deploy the tiered-memory system in your harness to enable persistent multi-tier context across resets.

Dependency Matrix

Required Modules

None required

Components

Standard package

💻 Claude Code Installation

Recommended: Let Claude install automatically. Simply copy and paste the text below to Claude Code.

Please help me install this Skill:
Name: tiered-memory
Download link: https://github.com/Cheggin/request-for-startups/archive/main.zip#tiered-memory

Please download this .zip file, extract it, and install it in the .claude/skills/ directory.
View Source Repository

Agent Skills Search Helper

Install a tiny helper to your Agent, search and equip skill from 471,000+ vetted skills library on demand.