The missing layer between your OS and agents.
0AL defines a runtime-agnostic execution model for securely running agents in zero-trust environments.
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 | 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 |
| SpecGraph | The executable ontology for Intent-Driven Development |
| SpecPM | The specifications registry for Intent-Driven Development |
Metrics | Metrics for AI-driven development |
Materials | Documentation, guides, and developer materials (planned) |
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.
To turn every process into a verifiable, minimal-trust agent — and make your OS agent-native.
We’ll be opening early issues, design discussions, and implementation proposals soon.
Until then — Trust No One. Run Everything.