Toward AI-Ready Auditing
A Model Context Protocol (MCP) Network for Oversight Institutions
Keywords:
Model Context Protocol (MCP), Artificial Intelligence Agents, Public Sector Auditing, Digital Government, Accountability, InteroperabilityAbstract
Public audit institutions face growing complexity in analyzing procurement, payroll, and budget data, while current AI applications remain fragmented and raise concerns about transparency and trust. This paper proposes a federated network of Model Context Protocol (MCP) servers, operated by oversight bodies to expose validated datasets and analytic workflows through standardized, policy-aware interfaces. AI agents interact with MCP servers under controlled authentication and logging, producing reproducible and auditable outputs. We outline a reference architecture, an explainable agent workflow, governance safeguards, and an illustrative use case in payroll auditing, arguing that MCP infrastructures can enable an AI-ready, citizen-centric audit ecosystem that enhances both efficiency and accountability.
References
DAWES, S. Interagency information sharing: Expected benefits, manageable risks. Journal of Policy Analysis and Management, v. 19, n. 2, p. 253–277, 2010.
JANSSEN, M.; KUK, G. Big and open linked data (BOLD) in government: A challenge to transparency and accountability. Government Information Quarterly, v. 33, n. 4, p. 563–567, 2016.
KARPAS, E. et al. What can AI agents do? A survey. Proceedings of AAAI, 2022.
OECD. Digital Government Review. Paris: OECD Publishing, 2020.
POWER, M. The Audit Society: Rituals of Verification. Oxford: Oxford University Press, 2021.
ZHANG, J.; YANG, L. Detecting anomalies in public sector data with machine learning. ACM Transactions on Management Information Systems, v. 12, n. 3, p. 1–18, 2021.
