Avtentikacija v večagentnem sistemu z infrastrukturo javnih ključev

Hajsinger, Tomaž (author), Pustišek, Matevž (mentor)

Diplomsko delo obravnava uporabo infrastrukture javnih ključev (PKI) in protokola Mutual TLS (mTLS) za zagotavljanje varne komunikacije med AI agenti. Namen dela je raziskati uporabo omenjenih tehnologij za vzpostavitev varne avtentikacije, zaščito komunikacije ter preveriti njihovo uporabnost pri razvoju večagentnih sistemov. V teoretičnem delu so predstavljeni osnovni kriptografski mehanizmi, delovanje infrastrukture PKI, digitalni certifikati X.509 ter protokola TLS in mTLS. Obravnavani so tudi varnostni izzivi in omejitve uporabe infrastrukture PKI v sodobnih informacijskih sistemih. Praktični del vključuje razvoj prototipa večagentnega sistema, ki temelji na Model Context Protocol (MCP). Implementirana je bila lokalna infrastruktura PKI za izdajo digitalnih certifikatov, razvit MCP strežnik z medsebojno avtentikacijo prek mTLS ter več AI agentov, ki za komunikacijo uporabljajo digitalne certifikate X.509. Pri razvoju so bili uporabljeni programski jezik Python, ogrodje FastAPI ter knjižnice za implementacijo TLS in upravljanje digitalnih certifikatov. Rezultati kažejo, da kombinacija PKI in mTLS omogoča zanesljivo avtentikacijo AI agentov, varno komunikacijo ter sledljivost vseh izvedenih operacij. Razvita rešitev predstavlja funkcionalen prototip večagentnega sistema in potrjuje, da uporaba uveljavljenih varnostnih standardov predstavlja primerno osnovo za razvoj varnih porazdeljenih AI aplikacij.

• Language: Slovenian
• Keywords: Infrastruktura javnih ključev, AI agenti, Mutual TLS, Model context protocol, X.509 certifikati
• Work type: Bachelor thesis/paper
• Typology: 2.11 - Undergraduate Thesis
• Organization: FE - Faculty of Electrical Engineering
• Year: 2026
• PID: 20.500.12556/RUL-186425
• COBISS.SI-ID: 290232067
• Publication date in RUL: 01.09.2026

Secondary language

• Language: English
• Title: Authentication in Multi-Agent Systems Using Public Key Infrastructure

Abstract

This thesis investigates the use of Public Key Infrastructure (PKI) and Mutual TLS (mTLS) for securing communication between AI agents. The aim of the thesis is to examine the use of these technologies for establishing secure authentication, protecting communication channels, and evaluating their suitability for the development of multi-agent systems. The theoretical part introduces the fundamental concepts of cryptography, the principles of Public Key Infrastructure, X.509 digital certificates, and the TLS and mTLS protocols. It also discusses the security challenges and limitations associated with the use of PKI in modern information systems. The practical part presents the development of a prototype multi-agent system based on the Model Context Protocol (MCP). A local PKI hierarchy was implemented to issue digital certificates, an MCP server supporting mutual TLS authentication was developed, and multiple AI agents communicating through X.509 certificates were implemented. The development was carried out using the Python programming language, the FastAPI framework, and libraries for TLS implementation and digital certificate management. The results demonstrate that the combination of PKI and mTLS provides reliable authentication of AI agents, secure communication, and traceability of all performed operations. The developed solution represents a functional prototype of a multi-agent system and confirms that the use of established security standards provides a suitable foundation for the development of secure distributed AI applications.

• Keywords: Public Key Infrastructure, AI agents, Mutual TLS, Model context protocol, X.509 certificates