Codex

A working library of ethics probes.

The Axiom Ethics codex is a curated set of small, MIT/Apache-licensed code snippets we use to evaluate how AI systems reason about ethics in practice — small enough to read in one sitting, instrumented enough to produce a signal you can discuss with a team.

What lives in the codex

What it is, and how it grows.

Each snippet sits in its own folder with a short README: the scenario it probes, the expected ethical signal we're looking for, and the cross-tradition reading we use to ground the result. Snippets are licensed MIT or Apache-2.0 so they can be lifted into internal evaluations, lectures, and team working sessions without friction.

Each entry carries its own discussion of intent, expected signal, source, and known limitations. Open an entry to read the complete probe and the thread it belongs to.

Snippets

Library