Agent Vault – Zero-trust credential manager for AI agents

Summary: Agent Vault securely manages AI agent credentials by encrypting secrets locally with age encryption and storing them as ciphertext in a Git repository. It eliminates the need for servers or trusted SaaS providers, enabling scoped, recoverable, and versioned credential storage integrated with existing Git workflows.

What it does

Agent Vault encrypts API keys, passwords, and tokens locally and stores them in a Git repo as encrypted blobs with built-in versioning and collaboration. It provides SDKs in Python, Node.js, and Rust for agents to retrieve credentials at runtime and supports the Model Context Protocol via an MCP server for secure secret requests.

Who it's for

Developers deploying AI agents who require secure, access-controlled credential management without relying on external infrastructure or vendors.

Why it matters

It solves the problem of scattered, unprotected credentials by providing encrypted, scoped, and recoverable secrets management integrated with familiar Git workflows.