Liquid Protocol: Bridging AI Agents and Solid Pods via the Model Context Protocol

Meem Arafat Manab, Víctor Rodríguez-Doncel, Beatriz Esteves

Keywords: AI agents, Solid Protocol, Model Context Protocol, GDPR compliance, data sovereignty, semantic web

TL;DR: Liquid Protocol enables AI agents to access user-controlled Solid Pods via MCP, achieving instant GDPR compliance and eliminating hallucination through structured RDF queries.

Abstract: We present Liquid Protocol, a specification for AI agent access to decentralized Solid Pods via the Model Context Protocol (MCP). Analogous to how the W3C Linked Data Platform (LDP) specified HTTP-based access to Linked Data resources for web applications, Liquid Protocol defines MCP-based operations for AI agents to access user-controlled RDF data. This architectural approach addresses two critical challenges: the inability to comply with GDPR's Right to Be Forgotten when personal data is embedded in model weights or vector databases, and the tendency of large language models to hallucinate information when lacking access to structured, authoritative data sources. Our reference implementation demonstrates instant Right to Be Forgotten compliance, elimination of hallucination through structured RDF queries, and preservation of data sovereignty through Solid's decentralized architecture.

Submission Number: 5