---
version: "0.1.2"
level: auto
processes:
  design: pair
  implementation: auto
  testing: auto
  documentation: copilot
  review: copilot
  deployment: assist
components:
  assets/: auto
  scripts/: auto
  tests/: auto
  rag/: auto
  knowledge/: copilot
  notes/: copilot
  presentations/: copilot
  AI-DECLARATION.md: copilot
  CLAUDE.md: pair
  AGENTS.md: pair
  fair.md: pair
  trust.md: pair
---

This format is based on [AI-DECLARATION.md](https://ai-declaration.md/en/0.1.2).

## Notes

- **This site is built by a human-directed AI agent pipeline, and says so.** Neuronautix
  writes about AI in research; declaring how its own site is made is the minimum
  consistency we owe readers.
- **Human accountability is unchanged.** Damien Huzard, PhD
  ([ORCID 0000-0003-4820-7951](https://orcid.org/0000-0003-4820-7951)) is the accountable
  author and owner of every published claim. `auto` describes *who typed it*, not *who
  answers for it*. Nothing reaches `main` without human review and merge.
- **`design: pair`** — site architecture, positioning, the three-pillar structure, the
  `fair.md` / `trust.md` conventions, and the editorial line are human-originated and
  argued out with the agent; the human understands the internals.
- **`implementation: auto`** — HTML, CSS, JavaScript, and Python are written by Claude
  (via Claude Code) running to completion from a prompt, under the agent roles defined in
  [`AGENTS.md`](AGENTS.md) and [`CLAUDE.md`](CLAUDE.md). Roughly half of this repository's
  commits are authored by the agent directly; most of the remainder are human merges of
  agent branches.
- **`testing: auto`** — everything under `tests/` and the CI checks in
  `.github/workflows/ci.yml` were agent-generated, then reviewed.
- **`documentation: copilot`** — the notes in `notes/` and the knowledge base in
  `knowledge/` are agent-drafted but gated: the publishing workflow has two mandatory
  human-in-the-loop approval gates (outline, then draft) before anything is published.
  Every source is retrieval-grounded; fabricated citations are not permitted.
- **`review: copilot`** — a `NoteReviewAgent` performs the first review pass and reports;
  it never approves. Human approval is required, and external review (Codex, Copilot,
  human) has produced corrections that are in the git history.
- **`deployment: assist`** — the GitHub Pages workflow and CI configuration are largely
  human-set with agent help.
- **Where to look first, if you are sceptical.** That is the point of this file:
  `assets/`, `scripts/`, `tests/`, and `rag/` are the fully generated parts. The
  prose in `notes/` is generated *and* gated *and* graded — per-claim epistemic
  categories and confidence scores are published in [`trust.md`](trust.md) and rendered
  inline by the site's Trust Lens.
- **This declaration is self-reported and not independently audited.** It is reviewed
  when the workflow changes; see `last_reviewed` below.

## Why this file exists

Neuronautix publishes on FAIR metadata, provenance, and AI agents in science. Two
repository-level manifests already sit at the root of this site:

- [`fair.md`](fair.md) — how FAIR this repository's knowledge artifacts are.
- [`trust.md`](trust.md) — how the knowledge was produced, how each claim is graded, and
  how confident the reader should be.

`AI-DECLARATION.md` is the third leg, and the most specific: not *how trustworthy is the
content*, but *which parts of this repository were written by a machine*. `trust.md`
already declares AI involvement in **authorship of knowledge**; this file declares it for
the **software and the site itself**, in a portable format that other repositories and
tooling can parse the same way.

The three are complementary and cross-referenced through their `companions` blocks. A
human-readable rendering of this declaration is published at
<https://neuronautix.com/ai-declaration/>.

## Scope

This declaration covers the `neuronautix.com` repository — the website, its knowledge
base, its build and QA scripts, and the local RAG prototype in `rag/`. It does **not**
cover client deliverables, consulting outputs, or the separate open-source projects
listed at </open-source-projects/>, each of which should carry its own declaration.

## Review

- `last_reviewed`: 2026-08-02
- Reviewed against: spec version 0.1.2
- Maintainer: Damien Huzard, PhD — Neuronautix, Montpellier, France