Build Production LLM Agents with Persistent Causal Memory
Sub-15ms semantic caching, pgvector HNSW hybrid retrieval, 2-hop Neo4j temporal graph traversal, and zero-trust E2EE envelope encryption.
import { MemoraClient } from '@memora/sdk';
// Initialize Memora Engine Client
const memora = new MemoraClient({ apiKey: process.env.MEMORA_API_KEY });
// Store & Query Hybrid Vector + Graph Memory
const result = await memora.search({ query: 'User preferred tech stack' });
console.log('Retrieved Memory Context:', result.memories);How Memora Powers Your AI Agents
3 seamless steps from SDK initialization to zero-trust memory retrieval.
Connect the SDK
Initialize with Python or TypeScript in under 3 lines of code using your self-service `memora_live_sk_...` key.
Store & Query Hybrid Graph
Memora automatically fuses vector embeddings with 2-hop Neo4j causal links and RediSearch semantic caching.
Zero-Trust E2EE Protection
AES-256-GCM envelope encryption isolates every workspace tenant with automated PII sanitization filters.
Hybrid Vector & Graph Search
Reciprocal Rank Fusion (RRF) combining vector embeddings, BM25 full-text, and 2-hop Neo4j causal links.
Envelope Encryption (E2EE)
Zero-trust workspace tenant isolation using AES-256-GCM envelope key derivation and PII sanitization filters.
Self-Service API Keys
Instant SHA-256 hashed API key minting (`memora_live_sk_...`) with full rate-limiting & telemetry scrapers.