Extended Reality (XR) is increasingly adopted for collaborative work, yet immersive storytelling and co-creation are often examined in isolation. This paper presents a conceptual review of XR systems where narrative structures and multi-user creation intersect. It introduces a four-dimensional framework encompassing Platform Affordances, Narrative Control, Participant Agency, and Adaptive Mediation. We position narrative not as an added layer but as a mechanism for organising collaboration. In this paper, we (i) synthesise how presence, interactivity, and authorship are operationalised across existing systems; (ii) define indicators for classifying XR co-creative experiences along the four dimensions; and (iii) outline directions for empirical validation, including comparative studies of control–agency trade-offs. We illustrate the framework by mapping representative platforms and sketching hybrid designs that balance authored coherence with emergent improvisation. Ethical considerations in AI-supported mediation, such as authorship, transparency, and accessibility, are also discussed. The framework offers a structure for comparative research and the design of XR systems where storytelling enables co-creative practice.