Skip to content

Memory Lifecycle

Varun Pratap Bhardwaj edited this page Mar 16, 2026 · 1 revision

Memory Lifecycle Management

SuperLocalMemory automatically manages memory freshness so your system stays fast and relevant without manual cleanup.

How It Works

Every memory moves through lifecycle states based on how often you use it:

State What It Means Your Memories
Active Recently used, immediately available Returned first in search results
Warm Used recently but not today Still searchable, slightly lower priority
Cold Haven't been used in a while Searchable but ranked lower
Archived Old and unused Compressed to save space, still recoverable
Tombstoned Marked for deletion Will be permanently removed

Automatic Transitions

You don't need to do anything — the system evaluates memories periodically and transitions them automatically. When you recall an archived memory, it's immediately promoted back to Active.

MCP Tools

Check Lifecycle Status

get_lifecycle_status

Shows the current state distribution of your memories.

Compact Memories

compact_memories

Triggers lifecycle evaluation and compaction. Use dry_run: true to preview what would change.

Bounded Growth

SuperLocalMemory enforces configurable storage limits. When your memory database approaches the limit, the oldest and least-used memories are archived or tombstoned first — ensuring your most valuable memories are always preserved.

What This Means For You

  • No manual cleanup — the system handles it
  • Fast search — only active/warm memories are prioritized
  • Nothing lost — archived memories can always be reactivated
  • Bounded storage — your database never grows unbounded

See also: Behavioral-Learning · Enterprise-Compliance · Upgrading-to-v2.8

Clone this wiki locally