Shipped, next, later.

• Materials authoring

  Material graphs, instances, parameter overrides, shading settings — with shader instruction counts on every compile.

• Animation Blueprints

  Anim graphs, full state machine authoring, and animation assets: montages, blend spaces, notifies.

• UMG Widget Blueprints

  Read and edit the Designer widget tree, wire widget events into the event graph.

• Plan mode

  Read-only exploration that ends in an implementation plan for your approval — before any edit.

• Atomic batch edits

  All-or-nothing edit batches: one failure rolls the whole batch back.

• Multi-model support

  Bring an OpenAI or Gemini key instead of (or alongside) Anthropic.

• Engine log access

  The agent reads the live editor session log — compile messages, Blueprint runtime errors, PIE output.

• Niagara

  Read and author Niagara systems and emitters.

• C++ editing with hot reload

  Edit native code and verify it through live coding, the same way Blueprint edits are compile-verified.

• MCP support

  Connect MCP servers so the agent can reach your studio’s internal tools, docs, and pipelines.

• Skills

  Teach the agent your project’s conventions and recurring workflows as reusable, on-demand skills.

Direction is firm; order may shuffle as we learn.

• Behavior Trees

  Author and edit AI Behavior Trees and Blackboards.

• Team collaboration

  Shared knowledge bases, team billing, and SSO for studios.

• Multi-agent systems

  Fan out subagents for large jobs — parallel refactors, project-wide audits, batch asset work.

• Perforce-native workflow

  Checkout-aware edits that respect your depot, not fight it.