AriadneMemory for AI agents

Local-first hybrid search + knowledge graph. Zero infrastructure.

Get Started

GitHub

Vector search (FAISS)

In-process FAISS index. Auto-upgrades from exact (Flat) to approximate (IVF) as your data grows.

Hybrid retrieval

Vector similarity + FTS5 keywords, fused with Reciprocal Rank Fusion.

Knowledge graph

Typed entities and relationships, multi-hop traversal. Edges are walked in both directions.

Cognitive retention

Ebbinghaus forgetting curve. Memories strengthen with use, fade without it.

Auto-dedup

MinHash LSH catches near-duplicates before they enter the store. Survives restarts.

Zero infrastructure

SQLite + FAISS, one .db file. No Docker, no Redis, no API keys, no daemon. Thread-safe.

Quick Start

python

from arriadne import AriadneMemory
from arriadne.embeddings import SentenceTransformerEmbedder

# An embedder turns text into vectors so semantic recall works automatically.
embedder = SentenceTransformerEmbedder("all-MiniLM-L6-v2")  # 384-dim

mem = AriadneMemory(db_path="memory.db", embedding_dim=embedder.dim, embedder=embedder)

# Store a memory
mem.remember("VPS has 4 cores, 8GB RAM, Ubuntu 24.04", importance=0.8)

# Search — vector + keyword, fused
results = mem.recall("server specs", k=5)
# → [{"content": "VPS has 4 cores...", "score": 0.94, ...}]

Capability	Ariadne	Chroma	sqlite-vec	Mem0
Vector search	✅	✅	✅	✅
Keyword + hybrid (RRF)	✅	⚠️	❌	⚠️
Knowledge graph	✅	❌	❌	⚠️
Near-dup dedup	✅	❌	❌	⚠️
Local, no daemon	✅	✅	✅	❌

_{Capability comparison, not a benchmark. ✅ built-in · ⚠️ partial/varies · ❌ not available.}

bash

pip install "ariadne-memory[embeddings]"