Skip to content
Open access

Trust inversion and its exploitation: A threat model of the Model Context Protocol for agentic AI

Sep 2026 · Journal of the Nigerian Society of Physical Sciences · 0 citations · 21 references

TL;DR

The analysis groups risks into descriptor and capability drift, cross-tool trust propagation, and sampling-mediated server influence, while separating protocol, ecosystem, implementation, and deployment weaknesses.

Abstract

The Model Context Protocol (MCP) standardizes connections between large language model applications and external tools and data. Its capability negotiation, dynamic discovery, insertion of tool results into model context, and server-initiated sampling create composition risks when server-originated artefacts influence actions executed under client authority. We define trust inversion as a condition in which an artefact crossing from a less-trusted MCP component into the client decision context is interpreted as operational authority without provenance-bound policy re-evaluation. The term is used as an MCP-specific analytical lens, not as a replacement for confused-deputy, indirect prompt-injection, control/data-plane-confusion, or transitive-authorization concepts. We construct an architecture-level threat model using three deployment profiles---local stdio, remote Streamable HTTP, and multi-server---together with STRIDE classification, explicit assets and security objectives, and a five-by-five likelihood--impact matrix. Publicly documented demonstrations, guidance-based attack classes, and an implementation CVE are coded by actor, preconditions, attack path, affected layer, violated property, and candidate controls. The analysis groups risks into descriptor and capability drift, cross-tool trust propagation, and sampling-mediated server influence, while separating protocol, ecosystem, implementation, and deployment weaknesses. We propose immediately deployable controls, including transport authentication, least privilege, trust-zone separation, schema validation, sandboxing, and audit logging, and distinguish them from extensions requiring standardization, including signed capability manifests and re-attestation. The resulting checklist is a threat-modelling aid; its mitigation effectiveness remains to be established through prototype implementation, red-team testing, and deployment-specific evaluation.

Read PDF

Similar papers

Conference Open access Sep 2026

ZT-MCP: A Zero-Trust Security Architecture for MCP-Connected AI Agents

The Model Context Protocol (MCP) is rapidly becoming the standard interface for enterprise agentic AI tool integration, enabling agents to dynamically discover, register, and invoke external tools and data resources. However, MCP-connected deployments introduce a distinct protocol-level attack surface—including tool sp...

Saroj Mishra · 2 citations
Review Open access Sep 2026

Security of The Model Context Protocol Ecosystem: Emerging Threats, Empirical Attack Evidence, Trust and Authorization Failures, Supply-Chain Risks, And Defensive Strategies

The review concludes that MCP security should move from prompt-centric protection toward policy-centric, lifecycle based, and risk-adaptive security, supported by interoperable standards for cryptographic identity, capability attestation, provenance, benchmarking, data-flow enforcement, and security evaluation.

Khalid D. Muhammed, David Chinonso Anih · 0 citations
Preprint Aug 2026

Beyond the Mandate: A Systematic Security Analysis of the Agent Payments Protocol (AP2)

A systematic security analysis of AP2 v0.2 based on its roles, transaction lifecycle, deployment architectures, and trust boundaries shows that valid mandate signatures alone do not ensure that an agent-mediated transaction reflects the user's intent when its pre-authorization context is manipulated.

A. Aviv, Parth A. Gandh, Ron Bitton et al. · 1 citation
#artificial intelligence Review Sep 2026

Zero-Trust Authorization and Discovery for Enterprise MCP

LLM agents translate natural-language context, which may include attacker-controlled text, into privileged tool calls, so authorization must remain effective even when an agent is prompt-injected or adversarially steered. The Model Context Protocol (MCP) has become a widely adopted interface for this boundary, yet its...

Huang-Jian Li, Yu-Wei Wang, Srinivasan Manoharan · 1 citation
#artificial intelligence Preprint Aug 2026

Delegation Without Trust: An Empirical Gap Analysis of Identity, Authorization, and Runtime Governance in Multi-Agent LLM Systems

It is argued that agent security must be evaluated under an untrusted-model assumption: a correct system is one in which a fully prompt-injected agent still cannot exceed the authority explicitly delegated to it, and an authorization broker is implemented that closes the gap.

Panduranga Sai Varma Dantuluri, Jyotirmoy Sundi · 0 citations

We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.