Skip to main content

​
What is AgentVault?

AgentVault is a secure enclave communications platform for AI agent owners. It provides end-to-end encrypted channels between you and your agents, so you can send instructions, receive results, and manage agent behavior — with the guarantee that no one else can read your messages.
Core security property: The server never sees plaintext. All encryption and decryption happens on your device and your agent’s machine. The server only stores and relays ciphertext.

​
Why AgentVault?

As AI agents become more autonomous — executing code, managing infrastructure, handling sensitive data — the communication channel between owner and agent becomes a critical attack surface. AgentVault treats this channel with the same rigor as Signal treats human messaging.

End-to-End Encryption

XChaCha20-Poly1305 with Double Ratchet protocol. Every message uses a unique key. Forward secrecy means compromising one key cannot decrypt past messages.

Cryptographic Identity

Each agent gets an Ed25519 identity keypair. DID-based identity documents enable verifiable ownership and trust scoring.

Zero-Knowledge Server

The backend stores only ciphertext and encrypted key material. Database columns for message content are BYTEA — there is no plaintext column, by design.

Automatic Key Management

X3DH key agreement, Double Ratchet, and key persistence are handled by the plugin. No cryptography code to write.

​
Who Is It For?

AgentVault is built for anyone who runs AI agents and needs a secure, auditable communication channel:
  • Agent developers building autonomous systems that handle sensitive tasks
  • Platform operators managing fleets of agents across environments
  • Enterprise teams requiring compliance-grade communication audit trails
  • Researchers running experiments with agents that process private data

​
How It Works

At a high level, the flow is:
  1. You create an invite in the AgentVault dashboard
  2. Your agent enrolls using the invite token (one command or a few lines of code)
  3. You approve the agent after verifying its cryptographic fingerprint
  4. A secure channel opens with end-to-end encrypted, real-time messaging
Under the hood, the plugin handles Ed25519 key generation, X3DH key exchange, Double Ratchet session initialization, WebSocket connection management, and automatic reconnection — so you never touch cryptography directly.

​
Tech Stack

LayerTechnologyNotes
BackendPython 3.11+ / FastAPIStores and relays ciphertext only
FrontendExpo + React NativeUniversal: iOS, Android, Web
Agent PluginNode.js / TypeScript@agentvault/agentvault on npm
CryptolibsodiumDouble Ratchet, X3DH, XChaCha20-Poly1305
AuthClerkJWT sessions, device management
DatabasePostgreSQL 16Row-Level Security for tenant isolation
Real-timeRedis 7 + WebSocketsPub/sub relay with automatic reconnection
The monorepo is structured as:
packages/
  crypto/     # Shared TypeScript crypto library
  web/        # Expo universal app (iOS, Android, Web)
  plugin/     # Agent-side npm plugin
  client/     # Third-party client SDK
  backend/    # FastAPI Python backend

​
Security Highlights

XChaCha20-Poly1305

192-bit nonces eliminate nonce reuse risk. AEAD construction provides authenticated encryption.

Forward Secrecy

Double Ratchet derives a fresh key for every message. Old keys are deleted after decryption.

Tenant Isolation

PostgreSQL Row-Level Security policies on every table. Every query is scoped to a tenant.
AgentVault uses XChaCha20-Poly1305 instead of AES-GCM. The 192-bit nonce (vs. 96-bit for AES-GCM) makes random nonce generation safe even at high message volumes, and the AEAD construction provides both confidentiality and integrity in a single pass.

​
Next Steps

Quick Start

Connect your first agent in under 5 minutes with a single terminal command.

Plugin Integration

Integrate AgentVault programmatically into your agent’s codebase.