Score every mesh, texture, material, and Blueprint in your project against production audit rules - naming, size, references, missing LODs, and invalid settings. Run the companion script in Unreal, drag the JSON here. Nothing uploads; the analysis never leaves this tab.
exec(open(r"C:/path/to/ta_asset_health_export.py").read())
download download ta_asset_health_export.py
| Asset type | Data collected (read-only) |
|---|---|
| Static Meshes | LOD-0 triangle count, LOD count, Nanite flag, material slots, referencer count |
| Textures | Dimensions, compression format, sRGB flag, mip setting, LOD group, NeverStream, referencers |
| Materials | Master vs instance, domain, blend mode, two-sided, dependency count, referencers |
| Blueprints | Parent class, tick-enabled, component count, dependency count, referencers |
Written for UE 5.0-5.5 with defensive property access - properties it can't read on your engine version export as "unknown" rather than crashing. It never modifies assets.
Every asset starts at 100 and loses points per finding. The rules mirror the Scene Audit Checklist:
| Severity | Examples |
|---|---|
| CRITICAL | High-poly mesh with no LODs and Nanite off - non-power-of-two texture - normal map with sRGB on - Blueprint with 100+ dependencies - 8+ material slots |
| WARNING | Naming convention violations - 4K+ textures - missing mips - translucent / two-sided materials - tick-enabled Blueprints - NeverStream on large textures |
| INFO | Unreferenced assets - high component counts - heavily-used master materials worth auditing |
Grades: A >= 90 - B >= 75 - C >= 60 - D >= 40 - F below. Naming rules assume the standard SM_ / T_ / M_ / MI_ / BP_ prefixes from The Asset Pipeline.
This runs as plain HTML and JavaScript. The JSON you drop is parsed in this tab, held in memory, and never sent anywhere. Close the tab and it is gone. View source if you want to check.