Nowadays, eXtended Reality (XR) technologies are quite mature and affordable to users. However, to become actually adopted it iskey to provide end-users with easy support for creating content autonomously, without needing specific programming knowledge.Shifting the entire building process towards end-users is not feasible, as XR development requires various skills, usually involvinge.g. 3D modelling, code development, design. While end-users could configure an XR environment by using scene builders [23], [24],[8], they typically start struggling when defining its dynamic behaviour (i.e. how objects interact with users/other objects). Currently,tools supporting the articulation of dynamic behaviour are too complex for end-users, as they target skilled people [23], or limit tosupport 3D objects animations [17]. Also, existing software development life cycles involve only technical people and are too slow torespond to the evolving needs of variegated users' categories, also because professional developers often lack the domainknowledge to address user requirements. Thus, a viable way to make XR applications better comply with users' expectations is tohave users more active in the process, making them capable of 'programming' the intended behaviour.To solve this problem we propose an End-User Development (EUD) [11] methodology which involves three different roles:Element Builders (EB): professional developers, having good skills in 3D modelling and game programmingEnd-User Developers (EUDevs): users without specific programming skills but familiar with using computers and XR, whoneed to create or update XR contents for their work or leisureXR consumers: the final consumers of XR contentsEUDevs will start from pre-defined XR elements, provided by EB and distributed through web repositories; then, they will composeand configure these elements according to their needs. We will define a taxonomy providing a EUDev-understandable description ofsuch objects and of the actions they support. We will then give EUDevs an authoring tool for specifying the interactions amongobjects, resulting in the definition of the XR environment behaviour in the form of rules, readable and understandable as naturallanguage sentences. The EUDev will enter the rules while immersed in the XR environment through a dedicated interface. Besides,we will develop an intelligent Conversational agent that will guide EUDevs in translating their intention into correct rules and guidethem into fixing possible errors.Finally, we also validate the resulting system in three relevant application domains (virtual museum, smart home and training ) toshow the generality of the approach, and we will also provide empirical user experience evaluation for all the involved roles.