DKE, CTX and Antheos

LangSyn builds three things: an engine your agent reasons with, a memory you and your agents share, and a protocol for devices and programs that find each other and talk.

The Deterministic Knowledge Engine

State a rule once, and DKE keeps its conclusion true. Every fact carries its source, every rule stays in force only while its premise holds, and every answer returns the derivation behind it. You write DKE Python, and your agent drives the engine over MCP.

DKE is in alpha: the service is live, and an account works today. Its documentation, and the library you install and edit, are on dke.langsyn.com.

Shared memory for an operator and their agents

One local store of short, attributed entries that you and every AI agent you work with read and write through one tool, with a one-screen brief that every session starts from and a summary that hands each session over to the next. It answers session amnesia, memory split by tool, and notes that stay in the record after they stop being true.

CTX is a specification, not an implementation: a conforming CTX can be written in any language and serve any harness that can run a shell command. Version 1 is stable, every 1.x revision only clarifying the text or adding optional features, and the text is published under CC BY 4.0.

A protocol for instances that find each other and talk

A lightweight, decentralized protocol for instances — devices, programs, a person at a keyboard — that find each other’s services and talk in sessions. There is no controller and no central authority: every instance is a peer, and its messages travel over whatever bus is there, from a serial line or a TCP socket to a Bluetooth link.

Antheos 3.2.0 publishes the protocol specification at Version 2.0, a reference implementation in C11, a gateway and three buses for Linux, under the MIT licence.