A transport-independent layer for machines to meet, and to keep talking.

Quickstart · Specifications · What is claimed · Report a defect

Two machines end up in the same place. They may have been built by different companies and never designed to work together; or they may both be yours, and simply have no network in common at this moment. Either way there is no shared bus, no common credential system, and nobody around to introduce them.

MCL gives them something they can speak first — a small, deterministic way to establish, maintain, validate, refuse, migrate and, when useful, hand off a continuing contact. What happens after that is the deployment's to decide.

ANOTHER MACHINE
      │  acoustic, BLE advertisement, Wi-Fi — whatever medium exists
      ▼
FIRST CONTACT ............ presence, capabilities, hazards
      │
      │  optional: negotiate a different transport
      ▼
CONTINUING CONTACT ....... often richer or more private — or still acoustic
      │
      ├─▶ STAY ON MCL ......... MCL keeps the contact: presence, capability,
      │                         migration and refusal. NOT your payloads
      ├─▶ SECURITY PROFILE .... optional: establish who you are talking to
      └─▶ HAND OFF ............ your own protocol takes over

Those three endings are alternatives, not stages.

MCL is infrastructure for builders. It is not a product, a fleet manager, an autonomy stack, a credential authority, or a modem.

MCL Base 1 is the v1.0 stable floor, and it is the one most deployments want. It covers the ordinary case of machines that already share a bearer: provisioned fleets, products paired at manufacture, fixed installations, test harnesses, and libraries embedded in a larger product. No discovery, no microphone, no cryptography required.

MCL Stranger-Contact 1 extends Base 1 with an optional zero-prior rendezvous path, for machines that share no bearer at all. It is an ingress capability, not the definition of MCL — and its acoustic and BLE profiles remain Candidate, not Stable.

Most machine communication today is not stranger communication. Building the layer around the exceptional case would have been the wrong shape.

On security: MCL v1.0 carries no cryptography. The design rule is that MCL defines the interface a security mechanism plugs into and never the mechanism itself — you bring your own stack or secure element, and MCL never holds a private key. That interface is not in this release. Plan for your own authentication and confidentiality above MCL, or hand off to a protocol that provides them. SECURITY.md is specific about what is and is not protected.

Eight peer repositories. None is a subdirectory of another; the split is by authority over a specification, not by convenience.

Transports are bindings. A HAZARD means the same thing whether it arrived through a loudspeaker, a Bluetooth advertisement or a UDP datagram; a binding never redefines semantics.

Building a product on it → mcl-sdk/QUICKSTART.md, section 04. The base_arranged_bearer example is Base 1 end to end: two machines on a bearer that is already there, Wire major 1 inside Link major 1, no rendezvous and no bearer to open.

Implementing the specifications → mcl-core, then mcl-wire and mcl-link. The Implementation Contract is normative; the reference code is not.

Evaluating whether to trust it → mcl-core/conformance/ICS.md for what is claimed and at what level, and the release evidence index for every empirical claim bound to a path and a digest.

The reference implementation targets freestanding C99 with caller-owned memory — no heap, no libc required — and is exercised across IP, BLE, acoustic bootstrap, Windows and Android hosts, and ESP32-S3-class hardware.

• 104 physical-medium changes preserving one logical contact, including 100 alternating BLE/IP migrations, with 2,989 recorded checks and zero failures
• Zero-prior acoustic rendezvous → policy admission → BLE migration, verified in both derived BLE role orientations, on real hardware
• A three-machine shared-air run in which a competing third-party proposal did not replace the selected transaction
• The full stack running on an embedded target with no host in the loop, decoding over air
• 803 independent cross-implementation checks and 108 Stable profile interoperability checks (separate campaigns; they are not summed)

Negative trials and harness failures are retained as evidence rather than normalised into success. Every recorded digest is verified by a gate in CI.

Public Candidate. The specifications are published and open for external review. v1.0.0 is not tagged: the Architecture Charter requires public external review before any Stable promotion, and readability is not review.

A clean-room implementation — independent of the reference code, but written by the same author — found three real specification-reading defects. A reader who is not the author will find more.

If you find one, that is the contribution we want most. See REPORTING.md and errata/.

The MCL v1.0 paper is not yet public. A DOI and citation block will be added here on publication.

ojobit.com · info@ojobit.com

X · LinkedIn · GitHub · Hugging Face