What problem does it solve? Building procedural content in Unreal Engine 5.8 requires choosing the right generation mode, data contract, and ownership model, and mistakes lead to non-deterministic output, runtime pop-in, broken HLODs, or GPU paths that are slower than CPU. This Skill guides the full PCG workflow so graphs are deterministic, profiled, and correctly integrated with World Partition, Data Layers, and HLOD. ## Core Features & Use Cases - Generation-mode selection: Chooses between editor, non-partitioned, partitioned, hierarchical, and runtime generation based on domain size, persistence, and streaming needs. - Graph design and data contracts: Defines spatial sources, metadata attributes, seeds, exclusion rules, and output ownership before building the graph in dependency order. - Maturity and platform guardrails: Enforces UE 5.8 constraints for Experimental PCG Editor Mode, manual Data Overrides, Beta GPU Processing, and Experimental GPU Procedural Instancing. - Use Case: Scatter vegetation across a large open world by sampling landscape surfaces, filtering slopes and road exclusions, driving species selection from attribute tables, spawning instanced meshes on hierarchical grids, and validating determinism across ten seeds. ## Quick Start Use the unreal-pcg skill to design a deterministic partitioned vegetation scatter graph for my Unreal Engine 5.8 landscape with road exclusion zones and HLOD support.