No user-submitted related servers found.
Scored across 305 tools
Multiple near-duplicate clusters exist: three tools invoke any Blender operator (capabilities-invoke, rna-call_operator, ui-operator_invoke), four search the API (capabilities-search, api-search, rna-search, topbar-command_search), and property access is split confusingly between properties-get/set and rna-get_property/set_property. Object creation and transform are also duplicated (object-create vs scene-create_object; object-transform_set vs scene-set_transform). An agent must read carefully to avoid misselection.
The vast majority of tools follow a consistent domain-verb_noun snake_case pattern (object-create, mesh-bevel, modifiers-add, uv-pack_islands), making the set predictable. Minor deviations exist: asset_class- uses an underscore in the domain, tool-settings vs tool-setting_get/set mixes plural and singular, and a handful of report tools use bare nouns (anim-timeline, uv-layers, sequencer-modifiers) instead of verb_noun.
305 tools is an extreme count that far exceeds the 50+ threshold for a mismatch. Even with strong domain organization, the sheer volume creates enormous selection overhead and makes it impractical for an agent to audit or reliably choose from the surface. Blender's breadth does not justify this many curated entry points when generic invoke/search tools already exist.
The curated surface is very broad: mesh editing, modifiers, materials, UV, rigging, animation, particles, physics, rendering, compositing, import/export, and scripting are all covered with mostly complete lifecycles. However, notable Blender workflows are absent from the curated set — sculpting, texture painting, object join/separate, NLA/drivers, and armature bone deletion — with generic invoke-any-operator tools papering over those gaps rather than filling them.