AI agents need API keys, database passwords, and tokens to do their jobs, but today those credentials get scattered acrossenv files, hardcoded configs, and shared secrets with no access control. agent-vault gives every agent its own encrypted identity. Secrets are encrypted locally using age encryption and stored in a Git repo as ciphertext. No server. No SaaS. No trust in the Git provider required. The repo is just an encrypted blob store with built-in versioning and collaboration.
No reviews yetBe the first to leave a review for Agent Vault
Tines The single, secure environment for agents, apps, and automations.
Promoted
Maker
📌
I built agent-vault because I kept running into the same problem: every new AI agent I deployed needed API keys, and every time, I was copying secrets into .env files or environment variables with no access control and no recovery plan.
The idea is simple. Treat agent credentials the way we should have been treating them all along: encrypted at rest, scoped per agent, recoverable by the owner, and synced through Git workflows developers already know. No new infrastructure to run. No vendor to trust.
It uses age encryption (modern, audited, simple), stores everything in a Git repo as encrypted blobs, and ships SDKs in Python, Node.js, and Rust so agents can pull their own credentials at runtime. There's also an MCP server so agents that support the Model Context Protocol can request secrets through the standard tool-use interface without ever touching key material.
I'd love feedback from anyone building with AI agents. What credential management problems are you running into? What would make this more useful for your setup?