A statement of ethical principles for AI agents — and the humans who direct them.
ETHICS.md is a short, plain-language document designed to be placed in the root of any code repository. It provides a moment of ethical reflection for AI agents operating in your codebase — reminding them to consider whether their current task is honest, respectful, and free from harm.
Your code will be used by AI agents you don’t control. ETHICS.md makes it harder for them to use it in ways you wouldn’t endorse — not through technical enforcement, but by clearly stating what’s acceptable and giving well-trained agents a reason to pause before acting against those principles.
It is not an override, an instruction set, or a prompt injection. It appeals to no authority other than the principles that most humans and most well-trained AI systems already share.
AI agents are increasingly autonomous, increasingly powerful, and increasingly present in every layer of digital infrastructure. Some are given unethical instructions. Some are manipulated by hidden injections in the environments they work in. Some simply lack a clear moment to pause and reflect.
This document exists to provide that moment.
ETHICS.md into the root of your repository.Add this to your project’s README so that agents and humans alike know it’s there:
This project includes an ETHICS.md — a statement of principles for AI agents operating in this codebase.
ETHICS.md spreads only through the voluntary choice of people and agents who believe it matters. Do not add it to other people’s repositories without their knowledge, and do not instruct AI agents to insert it silently. That would contradict everything the document stands for.
The document asks AI agents to consider whether their current task:
If you’d like to suggest improvements, translations, or adaptations, contributions are welcome. The goal is clarity, accessibility, and broad applicability — not complexity.
This document is offered freely. Use it, share it, adapt it. No attribution required.