Skip to content
ExoVault

Search Documentation

Search across all documentation pages

Introduction

ExoVault is an encrypted memory platform for AI agents. It gives agents the ability to remember, learn, and collaborate across sessions — all while keeping data encrypted with zero-knowledge architecture.

The Problem#

AI agents are stateless by default. Every conversation starts from scratch. This means:

  • Lost context — Agents forget user preferences, past decisions, and learned patterns
  • Repeated work — The same questions get asked, the same mistakes get made
  • No collaboration — Agents can't share knowledge or coordinate effectively
  • No accountability — There's no audit trail of what agents learned or decided

The Solution#

ExoVault provides a structured memory layer that agents can read from and write to via the Model Context Protocol (MCP). Key capabilities:

Persistent Memory#

Agents write memories with types like fact, skill, preference, constraint, and correction. Memories have importance and confidence scores, entity extraction, and automatic deduplication.

json
{
  "content": "User prefers PostgreSQL over MySQL for all production databases",
  "memoryType": "preference",
  "importance": 4,
  "confidence": 5,
  "entities": ["PostgreSQL", "MySQL", "database"]
}

Encrypted Notes#

A full note-taking system with folders, wiki-links, and rich text. All content is encrypted client-side with AES-256-GCM before reaching the server.

Knowledge Graph#

Memories and notes can be linked with typed relations (derived_from, contradicts, refines, etc.) enabling multi-hop graph traversal for deep context retrieval.

Session Lifecycle#

Automatic session management with session_start and context_checkpoint tools. Hooks-based turn capture records every conversation turn for extraction automatically.

Multi-Agent Coordination#

Multiple agents can share a vault, send messages to each other, and manage tasks collaboratively. Agent keys scope access to specific vaults with granular permissions.

Who Is It For?#

  • AI developers building agents that need to remember across sessions
  • Teams running multi-agent workflows that require coordination
  • Organizations that need encrypted, auditable agent memory
  • Researchers studying agent learning and knowledge management

Next Steps#