AxonOS

An open operating layer for brain–computer interfaces

The concept  ·  Core thesis  ·  Architecture  ·  The stack  ·  Evidence  ·  Reading order  ·  Standard ↗

---

The concept

▶ Watch the full concept video  ·  1:05

_{A simulated concept demonstration built from EEG-style telemetry and gameplay input.
It is not a clinical result, not a medical-device demonstration, and not a regulatory claim.}

---

What AxonOS is

AxonOS is a pre-clinical technical infrastructure project for the brain–computer interface boundary. It defines the operating layer that sits between neural acquisition hardware, real-time signal processing, consent enforcement, typed neural-intent events, and the applications that act on them.

It is strictly an operating layer for the brain. It is not an AI-agent framework, not a chatbot runtime, not a generic Python SDK, and not a token project. Everything below the application — the timing guarantees, the neural-permission model, the consent state machine — is specified, openly licensed, and built to be independently verified.

The principle — applications should receive typed, consent-bound intent events, never unrestricted raw neural streams.

---

What this repository is

This repository — AxonOS-org/AxonOS — is the public entry point to the project. It is deliberately small. It carries the concept media and serves as the map to the canonical technical stack; it holds no safety-relevant implementation of its own.

The engineering lives in the dedicated repositories under The AxonOS stack. The canonical, normative specification is axonos-standard.

Read this repository as the project's front door — not as its implementation.

---

At a glance

Kernel language	Rust, #![no_std], #![forbid(unsafe_code)]
Formal verification	28 Kani harnesses — bounded model checking
Unsafe code	0 — forbidden crate-wide
Real-time target	ARM Cortex-M4F / M33
Worst-case response	≤ 1000 µs, proven (L1); 972 µs measured (L2)
Public repositories	7 — every layer of the stack, no private repositories
Engineering RFCs	8, in axonos-rfcs
Canonical standard	axonos-standard v1.0.0
Clinical engagement	1 MOU — ALS rehabilitation centre, north-eastern US
Evidence held	L1 (formally proven) + L2 (measured) — no L3 claimed

_{Each figure above is traceable to a repository in the stack. See Evidence discipline.}

---

The core thesis

Raw neural data should not become the default application interface.

An operating layer for a brain–computer interface must answer, before any application runs:

Question	Where AxonOS answers it
Which neural-derived data may cross the boundary?	axonos-standard — capability system
Which application is allowed to receive it?	axonos-sdk — typed-intent boundary
Which consent state authorised it?	axonos-consent — consent FSM
How long does the event remain valid?	axonos-standard — STANDARD.md
Is raw-signal access structurally prohibited?	axonos-standard — capability system
Is the event typed, auditable, and provenance-bound?	axonos-standard — STANDARD.md

AxonOS treats each of these as a contract to be specified and verified — not as an implementation detail left to each vendor.

---

The architecture

            neural hardware
                  │
                  ▼   acquisition boundary
        real-time kernel substrate          ← axonos-kernel
                  │
                  ▼   consent + neural-permission enforcement
        deterministic intent processing     ← axonos-consent
                  │
                  ▼   typed neural-intent events
            application SDK                 ← axonos-sdk
                  │
                  ▼
  assistive · research · intelligent applications

Every arrow is a contract. The Standard defines what must hold at each boundary; an implementation is free in everything else. The four-layer model and the three cross-cutting subsystems — consent, the Cognitive Hypervisor, and swarm coordination — are specified in STANDARD.md.

---

The AxonOS stack

Every repository below is public. There are no private repositories.

Repository	Role	Status
axonos-standard	Canonical technical standard and architecture manual	Canonical
axonos-rfcs	Engineering RFCs and design records	Normative when finalised
axonos-kernel	Real-time #![no_std] kernel substrate	Research-grade
axonos-sdk	Application-facing SDK and typed-intent boundary	Active
axonos-consent	Consent finite-state machine and neural-permission reference crate	Pre-clinical reference
axon-bci-gateway	Acquisition-boundary / OpenBCI hardware-in-the-loop gateway	Non-safety acquisition boundary
axonos-swarm	Distributed timing and coordination research	Experimental

Inside the canonical standard, four documents carry most of the weight: STANDARD.md · VALIDATION.md · CONFORMANCE.md · GOVERNANCE.md.

---

Evidence discipline

AxonOS grades every quantitative claim by the evidence behind it. The taxonomy is defined normatively in VALIDATION.md and summarised here.

Level	Meaning
L1	Formally proven — a machine-checked proof over the entire admissible input space.
L2	Measured — a measurement taken on the reference hardware under a stated protocol.
L3	Independently validated — reproduced by a party independent of the project.

The public repositories currently hold L1 and L2 evidence. AxonOS does not currently claim any L3 result. A claim's level travels with the claim — an L2 measurement is never presented as an L1 proof.

---

What AxonOS does not claim

AxonOS does not currently claim, and this repository must not be read as claiming:

• FDA clearance, CE marking, or medical-device approval in any jurisdiction;
• clinical efficacy, or independent clinical validation;
• certified medical-device status, or production-implant readiness;
• complete compliance with IEC 62304, ISO 14971, or ISO 13485.

These are possible future milestones. They are not present facts, and the project records them as such.

---

Where to start

Recommended reading order

1. axonos-standard — STANDARD.md — the canonical specification.
2. axonos-standard — VALIDATION.md — the evidence discipline.
3. axonos-standard — CONFORMANCE.md — how an implementation is tested.
4. axonos-standard — GOVERNANCE.md — how the Standard evolves.
5. axonos-consent — the consent finite-state machine in #![no_std] Rust.
6. axonos-sdk — the typed-intent application boundary.
7. axonos-kernel — the real-time kernel substrate.
8. axon-bci-gateway — the acquisition-boundary gateway.

Frequently asked

Is this a product I can install? No. This is the project's entry-point repository. The engineering lives in the stack; the specification is axonos-standard.

Is AxonOS a medical device? No. See What AxonOS does not claim. A device built on a conformant implementation remains subject to the full regulatory regime of every jurisdiction in which it is deployed.

Can I build a conformant implementation? Yes — that is the purpose of an open standard. You need STANDARD.md and CONFORMANCE.md. An independent implementation that passes the conformance suite is exactly as conformant as the reference implementation.

Why is this repository mostly empty? Because it is a front door, not a workshop. A namesake repository crowded with implementation would obscure the map. The work is in the stack.

---

The project beyond the code

Website	axonos.org
Technology	axonos.org/technology.html
Specifications	axonos.org/specifications.html
Research & preprint	axonos.org/research.html · preprint
Engineering memo	axonos.org/memo.html
Standards & governance	axonos.org/standards.html · governance
Roadmap	axonos.org/roadmap.html

---

Contact

axonos.org  ·  github.com/AxonOS-org

General — connect@axonos.org  ·  Security — security@axonos.org

_{Released under the MIT License. Repositories in the stack state their own licences — the canonical Standard is released under CC-BY-SA-4.0.
© 2026 The AxonOS Project · Singapore}