List render and task capabilities

The published workflow corpus keyed by skill name: its description, verification status, Markdown content and content hash.

Full JSON Schema for every field each element type accepts, keyed by element type value (e.g. 'text', 'video') — everything element_types doesn't cover (accepted fields, types, defaults, constraints, enum options), including nested sub-objects like motion/effects/word_animation inlined under each entry's own $defs. Generated directly from the same Pydantic models the API validates requests against, so it can't drift out of sync.

Full JSON Schema of the render request body itself — the composition-level surface element_schemas doesn't cover: groups (transform groups, nesting, stack layout), custom_animations (inline keyframe animation registries), motion_blur, background_gradient, transitions, canvas and output settings. The elements array's per-type schemas are NOT duplicated here — read them from element_schemas. Generated from the same Pydantic model the API validates against, so it can't drift out of sync.

Self-contained JSON Schema per block, keyed by block name (e.g. 'geometry', a shape element's geometry): each carries only the $defs it reaches, so one block can be read without its element schema. Generated from the same Pydantic models the API validates against.

The property registry an operation's property names (render-node's, vendored and hash-checked): properties maps each path to its value type, unit, range, clamp, phase, composites, element kinds and drawn kinds; aliases maps alias spellings to paths; lint_codes names the path diagnostics; fnv1a64 is the registry's hash.