Most of the time you lose in the editor is not spent solving anything. What an AI agent covers, the read-back loop that separates working from guessing, and the work where it genuinely does not help.
Loading…
Drop the plugin into your project, point your MCP client at it, and let your AI actually build inside the editor.
Most of the time you lose in the Unreal editor is not spent solving anything. It goes on the Blueprint graph you have wired forty times, the material you could describe faster than you can build, renaming a hundred assets and then fixing the redirectors, and clicking through the same six panels to check whether a change did what you expected.
An AI agent for Unreal is a way to hand that half over. Not code you paste in afterwards. It calls real editor commands, and the asset appears in your content browser while you watch. Here is what that actually covers, and where it does and does not pay off.
Writing is the easy half. Anything can generate a plausible-looking Blueprint. The difference is whether it can look at the result.
Compile a material and list the errors that came back. Read a Niagara stack and see which stage a module actually landed in. Run a rig over an animation and get told which bones moved. Take a screenshot of the viewport and judge the result from it.
That loop is the whole thing. An assistant that can only write is guessing. One that can compile, read the result back and correct itself is doing the job. Every capability list in this space is really a proxy for whether that loop closes.
This matters more than it sounds. A tool that covers half the editor sends you back to clicking for the other half, and a workflow with a hole in it is not really a workflow. The hole is where the time goes back.
CodeFizz Editor Agent exposes 900+ commands across 28 categories, all of it editor side and all of it callable from a prompt:
That count is generated from the CLI source rather than estimated, because it has been wrong in several places at once before and I got tired of correcting it. Every command is searchable on the tools page if you want to check something specific is there before committing to anything.
Work you have done before. The forty-first Blueprint graph of a shape you have built forty times. Describing it is genuinely faster than wiring it, and the result is consistent instead of subtly different depending on the day.
Work that is mechanical but fiddly. Rig control hierarchies, gizmo sizes measured from the mesh rather than dragged until they look right, pole vectors placed from the limb's own geometry. Arithmetic dressed up as art direction.
Work with a verify step. Profiling is the one that changes how people work: run a trace, read which node cost the frame, change it, re-run to check it actually helped. That is a loop you would otherwise do by hand a dozen times an afternoon.
Work at volume. Forty characters that each need a competent rig and no more. One hero character you are art-directing is a different problem.
Taste. It builds the thing you described, not the thing you imagined, and the gap between those is your job. I have not once had it produce something I would ship untouched, and I would be suspicious of anyone claiming otherwise.
It is also not a substitute for knowing the engine. It will do what you ask precisely, which means asking for the wrong thing gets you the wrong thing quickly. The people who get the most out of it are the ones who already know what they want and are tired of clicking to get there.
Three commands and your project is reachable. The plugin runs a small bridge inside the editor, the CLI talks to it, and your assistant drives the CLI.
Install the CLI:
irm https://codefizz.dev/install.ps1 | iexInstall the plugin into your engine:
cfa install-plugin "E:/Unreal/UE_5.8"Check the editor is reachable:
cfa health_checkThen the step people skip and then regret:
Teach the assistant what the commands actually do:
cfa skill installWithout it the model guesses at parameter names. With it, it looks them up before calling anything. The difference in how often the first attempt works is not subtle.
It runs on Unreal Engine 5.6, 5.7 and 5.8, built and tested separately against each, because the engine ABI differs between versions and a single universal binary is not a thing that can exist.
For a real run with every command it called, including the part it got wrong, see using an AI assistant inside Unreal Engine 5. Setup in full is in the quick start, and the whole command surface is on the tools page.
Every claim above is something you can check on the tools page before paying anything. 900+ commands across 28 categories, on UE 5.6, 5.7 and 5.8. Drive it from Claude Code, Cursor, VS Code, or any MCP client.
Three day free trial. No card, nothing locked off, and it runs on your real project.
Want a corner of the editor reachable that is not covered yet? Say so in Discord. Most of what exists came from exactly that.