/agent-memory

Agent Memory

Four-tier memory architecture for learning agents

Bespoke TierUtility & Infrastructure

Overview

Scaffolds and manages tiered memory (session, working, consolidated, baseline) for multi-agent systems. Agents learn from sessions, incorporate feedback, and maintain context across hundreds of interactions.

What It Does

  • Initializes four-tier memory architecture (feedback, working, summaries, facts)
  • Captures user edits, ratings, and implicit signals into a feedback layer
  • Promotes old sessions from hot working memory to compressed summaries
  • Extracts queryable facts with provenance from session data via human-reviewable batches

Inputs

  • Project path
  • Agent count
  • Index dimensions
  • Optional workflow extension

Outputs

  • Memory directory structure
  • Session registry
  • Extraction prompts
  • Review queue batches

Example

/agent-memory

After 15 dialogue sessions with 5 wisdom-figure agents, run /agent-memory extract to pull effective formulations and emergent insights into a review queue. Approve the batch, and the evolved agent files gain new voice signatures.

Related Skills

/agent-coordination

Agent Coordination

Multi-agent setup, routing, and lifecycle management

/synthesize-knowledge

Knowledge Synthesizer

Extract and synthesize 10-100 documents at 10x speed

/memory-query

Memory Query

Query agent memory across all layers

Deep Dives

Knowledge OS Guide

Ready to use /agent-memory?

This skill ships with every Knowledge OS installation. Set up your system in 90 minutes.

View PricingAll 42 Skills

Built and maintained by Victor Sowers at STEEPWORKS