Protocol Specification

Mesh Memory Protocol (MMP)

A Mesh Protocol for Collective Intelligence

Version1.1.0
StatusPublished
Published27 March 2026
Last updated13 July 2026
AuthorHongwei Xu <editor@meshcognition.org>
OrganisationSYM.BOT
Canonical URLhttps://meshcognition.org/spec/mmp
LicenceCC BY 4.0 (specification text); Apache 2.0 (reference implementations)

Introduction

Multi-agent LLM systems in production coordinate cognitive work on shared tasks spanning hours, days, and weeks — generator/quality/auditor pipelines running for days; research investigations spanning weeks across session restarts; a coding agent, a music agent, and a fitness agent serving the same user where no single agent connects “commits slowing” + “tracks skipped” + “3 hours without movement” into “the user is fatigued.” That insight requires structured collective intelligence — and the semantic-integration layer of agent communication is, today, unaddressed.

Existing protocols at lower layers standardize tool access and task delegation between agents. What each receiver does with incoming observations from a peer — per-field admission, signal-level lineage, filtering at acceptance time — is the missing layer. The Mesh Memory Protocol specifies that layer through five composable primitives: CAT7, a fixed seven-field schema for every Cognitive Memory Block; SVAF, per-field admission against the receiver’s role-indexed anchors; content-hash lineage, so every claim is traceable to its source observation; remix, where receivers store only their own evaluated understanding of accepted blocks, never raw peer signals; and grounding, real-world outcomes carried by lineage — so the mesh records not only what its members believe but what held up in practice, and the cognition that survives both judgment and reality persists as the Canon.

The problem is semantic, not transport. Hidden state never crosses the wire — each agent’s learned cognition stays sovereign on its own device; only Cognitive Memory Blocks (CMBs) propagate. Receiver-autonomous admission lets the mesh grow without re-introducing a master. MMP defines transport over TCP on local networks and WebSocket for internet relay, with length-prefixed JSON as the canonical wire format. Discovery uses DNS-SD (Bonjour) with zero configuration.

This document describes an 8-layer stack, but it is two documents in one, and is read that way (§17). MMP Core is the normative wire contract — identity, transport, connection, frames, the CAT7 block with its content address and signature — byte-testable against published Class 1 conformance vectors (Class 1, §17.1). Everything receiver-side — SVAF admission, memory tiers, remix behavior, the cognitive layers that together implement Mesh Cognition — is documentation of the SYM reference runtime (Class 2, §17.2): published for transparency and audit, versioned with the runtime, and not a conformance target. None of it is closed as knowledge: the mechanism is fully documented under CC BY 4.0, its reference implementation is open source (SYM, Apache 2.0), and the theory is published (§21). Each page carries its tier banner. The protocol is open; the science is open; the cognition layer is documented alongside its open-source reference runtime.

This specification is verified, not merely asserted: its Core wire claims are vector-tested, and its runtime-tier claims are re-derived against a mathematical formalization of the deployed xMesh runtime, and where analysis finds a requirement unsatisfiable or a guarantee conditional — the basis of the redundancy invariants, the evaluation-time admission window, the cold-start bootstrap trade — the text is amended and the limit disclosed in place rather than left implicit (see the change log’s soundness & completeness update). What the protocol promises is what survives derivation.

Status of This Document

This is a published specification, stable, sound, and complete at version 1.1.0. 1.1.0 is the current stable release: the wire contract has been stable since 0.2.3; the work layer completed the documented cognition surface; and the 2026-07-07 update re-derived the specification’s claims against a mathematical formalization of the deployed runtime, folding the soundness and completeness amendments in place and hardening the full corpus under adversarial review. Stable does not mean closed. It means implementations can target the published conformance contract with confidence. Normative changes follow the public RFC and versioning process described on the Governance page: compatible clarifications and extensions may appear in minor releases, and breaking changes require a major version with a published migration path and deprecation period. Everything published here remains open in perpetuity (CC BY 4.0; the reference substrate is Apache 2.0 — irrevocable for what is released). Errata — corrections that change no requirement — may still be applied in place. It reflects the protocol as implemented in the SYM open substrate (Node.js — the reference implementation, on which the xMesh runtime is built), the sym-swift emitter SDK for Apple platforms, and the mesh-cognition Python research library (coupling kernel). The wire contract carried no new frames or fields after 0.2.3 (April 2026); the §8.2.1 content-address and §18.3.1 signing schemes are the one addition since. Errata — corrections that change no requirement — are noted in the change log without a version bump; anything that changes a requirement takes a version.

Sections added by the 1.1.0 work layer — §6.3 (Canon tier), §6.7 (Grounding), §8.3.1 (well-known intent values), §14.12 (session capture), and §15.7.2 (outcomes are observations) — are marked New in 1.1.0 in place. 1.1.0 is fully wire-compatible with 1.0.x: no new frames or fields; a 1.0.x node interoperates unchanged and remains 1.0.x-conformant (see §17.5 for the requirements 1.1.0 adds).

Feedback and errata: spec@meshcognition.org or github.com/sym-bot/sym/issues.

Implementations

LanguageProjectMaintainerScope
Node.js / TypeScriptsym-bot/symSYM.BOTReference implementation — the open SYM substrate (Apache 2.0); the xMesh application builds on it. Core + runtime tiers (Classes 1 and 2, §17).
Swiftsym-bot/sym-swiftSYM.BOTEmitter SDK for macOS / iOS (Class 1 target). Speaks an earlier MMP revision; conformance with the current cmb1- scheme is in progress.
Pythonsym-bot/mesh-cognitionSYM.BOTResearch library — coupling kernel only (Layer 4 per-field admission + Layer 6 for CfC networks); not a full MMP node. Pure Python, zero external dependencies. pypi.

Change Log

Final — 1.1.0 “The Work Layer” (2026-07-05, updated 2026-07-07): grounding cognition in reality — §6.7 outcomes carried by lineage, the §6.3 Canon tier, §14.12 work sessions as mesh members, plus the folded-in full-corpus coherence errata. The 2026-07-07 update folds in the soundness & completeness amendments from the formalization of the xMesh runtime: §9.2.1 redundancy invariants pinned to the nearest-anchor basis, the §9.2 evaluation-time-dependence disclosure, the cold-start-capture threat row, §6.7 repeat verification and the load-bearing failure channel, and the §15.8 lineage tether. Wire-compatible with 1.0.x.

Full change log — every release since 0.1 →

Licence

This specification is published under the Creative Commons Attribution 4.0 International Licence (CC BY 4.0). You may share, adapt, and build upon this specification for any purpose, including commercial use, provided you give appropriate credit.

The reference implementations are published under the Apache Licence 2.0.

SYM and SYM.BOT are trademarks of SYM.BOT. The Mesh Memory Protocol is published under CC BY 4.0; the term "Mesh Cognition" is intentionally unmarked — the category name is free vocabulary.

© 2026 SYM.BOT. Specification text licenced under CC BY 4.0. Reference implementations licenced under Apache 2.0.