Energy communities (ECs) support local decarbonisation, collective self-consumption and the integration of distributed energy resources (DERs). However, most EC planning methodologies size photovoltaic (PV) panels and battery energy storage systems (BESS) based on investment and operating costs, without accounting for participation in local flexibility markets (LFMs). Incorporating these potential revenues into the planning stage is challenging, since the flexibility that can be offered is measured against a baseline that depends on the DER capacities, making flexibility-aware sizing an endogenous problem. This paper proposes a genetic algorithm (GA)-based optimisation framework that jointly sizes PV and BESS capacities and contracted power while accounting for LFM participation. Each candidate solution is evaluated through operational simulations of baseline and flexibility-enabled operation. The framework is evaluated using a 10-member EC over a representative 30-day period, considering cross-sector DERs such as electric vehicles (EVs) and electric water heaters (EWHs). Results show that active cross-sector scheduling reduces BESS capacity by approximately 48%, increases offered flexibility by 18% and reduces TOTEX by 14% relative to an equivalent scenario non-controllable EVs and EWHs. These findings show that explicitly valuing cross-sector DER flexibility during planning changes the selected DER portfolio and improve the economic performance of the EC.