MSP-1 — Mark Semantic Protocol: An Open Semantic Clarity Layer for Agentic Systems

This document specifies MSP-1 (Mark Semantic Protocol), an open, voluntary, machine-readable clarity layer that enables publishers to declare semantic intent, authorship context, provenance, authority, and interpretive framing for web content. These declarations are intended for consumption by AI agents, answer engines, search systems, and other automated consumers as contextual signals when retrieving and interpreting that content. MSP-1 declarations are expressed as JSON-LD. Site-level declarations are served through a deterministic well-known discovery endpoint, while page-level declarations may be embedded within individual pages. MSP-1 creates no rights or obligations and defines no enforcement mechanism. This document specifies the MSP-1 discovery model, site-level and page-level declaration objects, core term vocabulary, conformance model, and guidance for interpreting MSP-1 declarations. This deposit is a versioned, citable snapshot of MSP-1 Version 1.0.2 as published on 2 September 2026. It is authoritative for the content of that archived snapshot. The current and continuously maintained specification is published at https://msp-1.org/. Version 1.0.2 updates the JSON-LD context reference architecture by introducing dedicated /context/ resources and separating JSON-LD semantic context from JSON Schema validation. It also harmonizes the specification, schemas, examples, and tooling with the established MSP-1 core vocabulary and declaration model, without introducing new core terms or changing the protocol's declarative, non-binding semantics.