name: memory-tiers
version: 0.1.0
description: Tiered memory management for AI agents — LRU cache for context. Hot/warm/cold tiers with automatic promotion, demotion, and access tracking.

Memory Tiers

A tiered memory system for AI agents. Like CPU cache hierarchy, but for your context window.

Architecture

MEMORY.md (Tier 1 — Hot)          ← Always loaded, ~100 lines max
  ↕ promote (accessed) / demote (stale 7+ days)
memory/tier2-recent.md (Warm)     ← On-demand, last 30 days
  ↕ promote / demote (stale 30+ days)
memory/tier3-archive.md (Cold)    ← Search only, deep history
  ↕
memory/YYYY-MM-DD.md (Raw)        ← Immutable daily logs

Scripts

All scripts are in scripts/ relative to this SKILL.md.

Track Memory Access

Scans session transcripts for memory file reads/writes/searches. Run regularly.

node scripts/track.js                    # scan last 24h
node scripts/track.js --since 72         # scan last 72 hours

Output: Updates state/access-log.json with per-file and per-section access data.

Run Maintenance

Promotes/demotes sections between tiers based on access patterns.

node scripts/maintain.js --dry-run       # preview changes
node scripts/maintain.js                 # apply changes
node scripts/maintain.js --demote-days-t1 5   # custom T1 demotion threshold

Rules:

• Tier 1 sections not accessed in 7 days → demote to Tier 2
• Tier 2 sections not accessed in 30 days → demote to Tier 3
• Tier 2/3 sections accessed in last 24h → promote to Tier 1

Manual Promote/Demote

Move a specific section between tiers.

node scripts/promote.js --list                           # show all sections
node scripts/promote.js --section "LARC" --from 3 --to 1  # promote
node scripts/promote.js --section "TextWeb" --from 1 --to 2 # demote

Health Report

Show current tier sizes, staleness, and recommendations.

node scripts/report.js                   # pretty-printed
node scripts/report.js --json            # machine-readable

Heartbeat Integration

Add to your heartbeat routine (every few hours):

## Memory Maintenance
1. Run `node <skill_dir>/scripts/track.js` to update access tracking
2. Run `node <skill_dir>/scripts/maintain.js --dry-run` to check for needed changes
3. If changes look good, run without --dry-run
4. Run `node <skill_dir>/scripts/report.js` to verify health

State Files

• state/access-log.json — Per-file and per-section access timestamps and counts
• state/maintenance-log.json — History of maintenance actions taken

Tier Rules

| Rule | Threshold | Action |
|------|-----------|--------|
| T1 stale | Not accessed 7 days | Demote to T2 |
| T2 stale | Not accessed 30 days | Demote to T3 |
| T2/T3 accessed | Accessed in last 24h | Promote to T1 |
| T1 over limit | >100 lines | Demote lowest-access sections |
| Daily log | Never modified after day ends | Source of truth |

File Conventions

Tier files should have these headers:

# MEMORY.md — Working Memory (Tier 1)
<!-- last_maintained: YYYY-MM-DD -->
<!-- tier: 1 — hot context -->

Sections use ## or ### headers. The maintenance scripts parse these to identify movable units.

The ## Index → Deeper Tiers section in MEMORY.md should always be last — it provides pointers to content in lower tiers.