Systems language · Normative work in progress

One language.
Five ways to reason.

ZergLang makes scheduling, authority, ownership, and evidence part of the program—whether code stays interpreted or becomes native.

Start exploring Inspect the evidence

Core-0 targetCore-1 previewStable Core-1 inactive

signal.zlchecked

module signal;

public value Signal {
    public compiled algorithm message score(type: Class<Self>,
        input: Int64) -> Int64 effects {} {
        return input * input;
    }

    public interpreted algorithm message explain(type: Class<Self>,
        item: Dynamic) -> Dynamic effects { reflect } {
        return item;
    }
}

Stage0Working bounded algorithm slice

ZLM1 + ZLM2Versioned checked artifacts

ZL256Task-level evidence, gaps included

The domain model

Different work deserves different laws.

The five domains share objects, contracts, errors, and ordinary expressions. Each narrows scheduling and effects instead of creating a separate language.

algorithm

Ordered computation, contracts, ownership, and ordinary control flow.

compute

Checked tensor dataflow, shapes, strict numerics, and explicit devices.

state

Serializable transactions, deterministic reactions, and versioned cells.

flow

Recorded activities, durable checkpoints, retries, and replay-safe history.

optimize

Finite search, hard acceptance, explicit budgets, and evidence-bound reports.

Authoring modality

Explore in the open. Compile the proven closure.

Interpreted code can remain open-ended. Compiled code accepts a stricter materializable contract. An agent may propose either conversion, but admission still requires checked identities and behavioral replay.

Read how conversion works

interpretedDynamic · exploratory

validated rewrite

compiledclosed · materializable

one checked module · stable contract ID

Inspect, don’t infer