Hystersis

Hystersis is an open-source long-term memory layer for AI agents. It provides persistent memory, semantic search, knowledge graphs, and real-time multi-agent sync.

Why Hystersis?

AI agents forget everything between sessions. Hystersis gives them persistent memory that adapts, compresses, and improves over time.

Key Capabilities

Semantic Memory — Store and retrieve by meaning, not keywords
Knowledge Graph — Neo4j-powered entity relationships with 30+ edge types
Spreading Activation — Multi-hop reasoning with +23% accuracy
ProMem Compression — 80-85% token reduction at 97% accuracy
Temporal Reasoning — Phase rotation preserves history
Memory Worth Scoring — Outcome-linked importance
Safety Classifier — Blocks prompt injection and PII

Quick Start

docker compose up -d
curl -X POST http://localhost:8080/memories \
  -H "Content-Type: application/json" \
  -H "X-API-Key: your-key" \
  -d '{"content": "User prefers dark mode", "type": "user", "user_id": "alice"}'