dmno-dev/varlock

AI-safe .env files: Schemas for agents, Secrets for humans.

What it solves

Varlock solves the problem of managing environment variables and secrets in a way that is safe for AI agents. Traditional .env files often lead to secret leaks or lack of-sync synchronization between example files and actual values. Varlock introduces a schema-based approach where AI agents can understand the configuration requirements (types, validation, descriptions) without ever seeing the actual sensitive values.

How it works

It uses a new DSL called @env-spec to attach schemas and functionality to .env files using JSDoc-style comments. This allows developers to define a single source of truth in a .env.schema file. The tool provides a CLI for loading, validating, and running processes with resolved variables. It also features a plugin system to pull secrets from external backends like 1Password, AWS, Azure, and GCP, and includes a varlock scan command to detect leaked secrets in AI-generated code.

Who it’s for

Developers and teams building AI-integrated applications who need to secure their secrets while providing necessary context to AI agents and maintaining type safety across multiple environments.

Highlights

  • AI-Safe Config: Provides agents with variable names and types without exposing secrets.
  • Secret Leak Prevention: Proactive scanning via varlock scan and runtime protection including log redaction.
  • Type Safety: Offers validation, coercion, and IntelliSense support via a VSCode extension.
  • Extensible Backend Support: A wide array of plugins for major secret managers (e.g., HashiCorp Vault, Infisical, 1Password).
  • Multi-environment Management: Automatic loading of .env.* files and explicit imports.

Related

  • Project
    jingkaihe/matchlock

    Matchlock is an experimental CLI/SDK tool that runs AI agents inside fast-booting, disposable microVMs with network allowlisting, in-flight secret injection (real keys never enter the VM), no host filesystem sharing, and explicit file copy in/out. It supports Linux (KVM) and Apple Silicon macOS, offers Go/Python/TypeScript SDKs, and is MIT-licensed.

  • Project
    tophant-ai/ClawVault

    A security vault and proxy gateway that protects AI agents by detecting sensitive data leaks, blocking prompt injections, and intercepting dangerous system commands.

  • Project
    vercel-labs/deepsec

    An agent-powered vulnerability scanner for large-scale repositories that uses AI to find and verify deep-seated security issues.

  • Project
    tensorlakeai/tensorlake

    Tensorlake is a cloud platform that offers fast, scalable MicroVM sandboxes for running AI agents safely. It provides a Python SDK and CLI to create, manage, snapshot, and pool sandboxes, a versioned object‑store filesystem, and a serverless orchestration layer where functions (decorated with `@function`) run in isolated sandboxes and can be invoked via HTTP.

  • Project
    devchilll/scope

    SCOPE is a reference implementation of an AI agent governance framework for high-stakes applications, built on Google's Agent Development Kit (ADK). It enforces safety, compliance, access control, observability, and human escalation through five pillars, demonstrated with a banking customer service agent.