Enables AI agents to generate and compose thumbnails using tools like get_capabilities, measure, render_thumbnail, render_variants, and get_canvas, with a document model that supports layers, effects, and masking.
