AI Contributing Guide

AI systems may help draft, search, refactor, or test, but contributions enter through a human-submitted pull request. AI systems are tools or acknowledged assistants, not accountable authors.

A human accountable contributor may prepare the disclosure below and submit it through the public Animus Omega evidence archive issue tracker. Do not include private information. AI systems must not contact people, submit work, or claim authorization on their own.

Every submission must name one human accountable contributor. The pull request must include:

1. model name/version (if known), tool or service, and date;
2. a short prompt or instruction summary (do not include secrets);
3. the complete changed-file list;
4. test commands, results, and any unrun tests;
5. source and data provenance, including generated-material provenance;
6. license and attribution review, including any third-party text or media;
7. the human review statement and known limitations.

The named human verifies every factual claim, citation, dependency, license, and generated output. They must remove invented citations, unsupported precision, and unmarked placeholders. AI must not be used to fabricate sources, authorship, test results, or evidence. Do not paste credentials, private data, unpublished sensitive material, or other secrets into a model; report accidental exposure immediately and rotate the secret through the proper process.

AI-assisted code follows the same deterministic, manifest, review, and claim boundary requirements as any other code. An acknowledgment may say which system assisted and what it did; it does not transfer accountability or grant authorship.

An AI-assisted rerun, review, or implementation must not be described as an independent replication merely because it used a different model, agent, or prompt. Independent replication requires a separately accountable human or team, an independently controlled environment and execution, and disclosure of all AI assistance.