|Decarbonising and decontaminating remote regions in the world presents several challenges. Many of these regions feature isolation, dispersed demand in large areas, and a lack of economic resources that impede the development of robust and sustainable networks. Furthermore, isolated systems in the developing world are mostly based on diesel generation for electricity, and firewood and liquefied petroleum gas for heating, as these options do not require a significant infrastructure cost. In this context, we present a stochastic multi-energy multi-microgrid system planning model that integrates electricity, heat and hydrogen networks in isolated systems. The model is stochastic to capture uncertainty in renewable generation outputs, particularly hydro and wind, and thus design a multi-energy system proved secured against such uncertainty. The model also features two distinct constraints to limit the emissions of CO2 (for decarbonisation) and particulate matter (for decontamination), and incorporates firewood as a heating source. Moreover, given that the focus is on low-voltage networks, we introduce a fully linear AC power flow equations set, allowing the planning model to remain tractable. The model is applied to a real-world case study to design a multi-energy multi-microgrid system in an isolated region in Chilean Patagonia. In a case with a zero limit over direct CO2 emissions, the total system's cost increases by 34% with respect to an unconstrained case. In a case with a zero limit over particulate matter emissions, the total system's cost increases by 189%. Finally, although an absolute zero limit over both, particulate matter and direct CO2 emissions, leads to a total system's cost increase of 650%, important benefits in terms of decarbonisation and decontamination can be achieved at marginal cost increments.