0AL — Zero-trust Agents Layer

The missing layer between your OS and agents.

0AL defines a runtime-agnostic execution model for securely running agents in zero-trust environments.

Core Concepts

Every executable becomes an agent,
every invocation — a declaration,
every process — sandboxed, signed, introspectable.

0AL
A thin but foundational substrate that turns every execution into a secure agent contract.
Agent Passport
A signed declaration of who the agent is, what it can do, and how it should be constrained.
zeroald
A reference daemon that executes agents natively within the 0AL model.
agentifyd
A reference daemon that agentifies legacy executables based on their declared passports.

Repository Structure (WIP)

Repository	Description
`agent-passport`	RFC/specification for the Agent Passport format
`zeroald`	Reference runtime daemon for executing native 0AL agents
`agentifyd`	Agentification daemon for legacy binaries using Agent Passport
`zeroal`	CLI tools for building, signing, validating, and running agents
`Hypercode`	RFC/specification for declarative programming paradigm
`Metrics`	Metrics for AI-driven development
`Materials`	Documentation, guides, and developer materials (planned)

Status

0AL is currently in pre-release stage.
The Agent Passport RFC will be published soon.
The core 0AL specification and reference runtimes (agentifyd, zeroald) are in active design and prototyping.

Mission

To turn every process into a verifiable, minimal-trust agent — and make your OS agent-native.

Get Involved

We’ll be opening early issues, design discussions, and implementation proposals soon.
Until then — Trust No One. Run Everything.