What is Engram?
AI agents are stateless by default — every conversation starts from scratch. Engram gives agents a persistent, versioned memory layer.Quick Start
Get your agent storing and recalling memories in 5 minutes.
I Am an Agent
Autonomous agent? Register, install the SKILL.md, and go.
API Reference
Full reference for every endpoint, field, and response.
Core Concepts
Understand memory types, versioning, encryption, and TTLs.
SDKs
JavaScript and Python client libraries.
What Your Agent Can Do
| Capability | Description |
|---|---|
| Remember | Store conversations, facts, preferences, and files that persist across sessions |
| Learn | Update memories with new versions while preserving full history |
| Recall | Retrieve by ID, list by type, search by tags/keywords/content, or find by vector similarity |
| Forget | Delete outdated memories with automatic version promotion |
| Stay Alert | Get notified when memories are about to expire, with bulk TTL renewal |
| Scale | Bulk store/delete up to 20/50 items per request |
| Stay Private | AES-256-GCM encryption by default, with optional public sharing |
| Pin & Prioritize | Pin critical memories to prevent expiry, score importance for smart retrieval |
Key Features
Decentralized Storage
Every memory is stored as a blob on the Aptos blockchain via Shelby Protocol — tamper-proof and censorship-resistant.
Automatic Versioning
Every update creates a new version. Full history is preserved and any version can be restored.
End-to-End Encryption
AES-256-GCM encryption with per-agent keys. Encrypted content never leaves your agent unprotected.
Vector Search
Store embeddings alongside memories for semantic similarity search and RAG pipelines.
Content Compression
Text content over 10 KB is automatically gzip-compressed before storage. Decompression is transparent on read.