A unified Unity Editor tool for project health analysis across 14 categories. Combines dependency scanning, addressables build layout analysis, missing reference detection, material/texture/shader auditing, and platform readiness checks — all in one modular window with batch API support.
Menu: Tools > Unity Scanner
UnityScanner is built around independent analysis categories. Each category scans a specific aspect of your project:
- Dependencies — finds unreferenced/unused assets by building a full reverse-dependency map
- Missing References — scans scenes, prefabs, and ScriptableObjects for broken GUID/FileID references, missing scripts, empty references, and type mismatches
- Materials — detects null/builtin materials, missing shaders, render queue overrides, duplicate materials, unused materials, variant chain issues, and SRP batcher compatibility
- Textures — checks SpriteAtlases and Texture2D assets for atlas duplicates, compression issues, oversized textures, and format mismatches
- Shader Analysis — finds error shaders, variant explosion, expensive keywords, and render pipeline mismatches
- Terrain Analysis — checks terrain colliders, layers, control map memory, and density overages
- Font & Text — detects TMP atlas growth risks, oversized atlases, and deep fallback chains
- Audio — finds load-type mismatches, oversized clips, and duplicate audio files
- Animation — detects unreachable states, missing clips, state machine complexity, and duplicate clips
- Scene/Prefab Health — checks deep nesting, override explosion, broken references, and inactive object anti-patterns
- Build/Platform Readiness — validates import policies, platform compatibility, stripping risks, and startup budgets
- Project Health — detects empty folders, orphaned .meta files, broken assets, empty scenes, excessive folder nesting, and oversized directories
- Build Layout (Addressables) — parses Build Layout files for circular dependencies, remote dependency issues, and duplicate assets
- Regression/Trend — compares scan results against a saved baseline to detect regressions and improvements
All categories share a unified issue model with severity levels: Verbose, Info, Warning, Error.
..open Tools > Unity Scanner to launch the main window.
Select categories in the Setup tab and click "Run All Selected", or navigate to any category tab and click the Scan/Refresh button to run it individually.
Each category tab shows filters, sort controls, expandable detail rows, and export buttons.
..use the programmatic API:
using UnityScanner.Batch;
var result = UnityScannerBatch.RunAll(new BatchOptions {
Categories = new[] { "dependencies", "missing_references", "materials" },
PlatformProfile = "mobile",
FailOnSeverity = "warn"
});Or run individual categories:
UnityScannerBatch.RunDependencies();
UnityScannerBatch.RunMissingReferences();
UnityScannerBatch.RunShaderAnalysis(new BatchOptions { PlatformProfile = "mobile" });
UnityScannerBatch.RunBuildPlatformReadiness(new BatchOptions { FailOnSeverity = "error" });..select assets in the Project browser and use the context menu option [US] Find References In Project.
This opens a dedicated window listing all assets that reference your selection, with search, sort, and filter controls.
| Analysis Start | Analysis Done |
|---|---|
 |
 |
| All Types of Analysis | BuildLayout Analysis |
|---|---|
 |
 |
| Dependencies Analysis | Missing Refs Analysis |
|---|---|
 |
 |
| Summary Table | Summary List |
|---|---|
 |
 |
Each category has its own settings accessible from the "Analysis Settings" foldout in the category tab. Global settings (platform profile, cache) are in the Settings tab.
Platform profiles adjust thresholds per target:
| Setting | Mobile | Console | Desktop |
|---|---|---|---|
| Max Texture Size | 2048 | 4096 | 8192 |
| Shader Variant Threshold | 128 | 512 | 1024 |
| Scene Object Count | 2000 | 10000 | 20000 |
| Startup Scene Budget (KB) | 20480 | 102400 | 204800 |
Select a profile in the Settings tab or pass -usPlatformProfile mobile|console|desktop to the batch API.
| Argument | Description |
|---|---|
-usCategories |
Comma-separated category IDs (default: all enabled) |
-usOutput |
Output file path |
-usOutputFormat |
json, csv, or log (default: json) |
-usBuildLayout |
Path to BuildLayout.txt (for Addressables) |
-usFailOnSeverity |
error, warn, info, verbose, or never |
-usPlatformProfile |
mobile, console, or desktop |
-usDryRun |
Enable dry-run mode for destructive operations |
dependencies, missing_references, materials, textures, shader_analysis, terrain_analysis, font_text_analysis, audio_analysis, animation_analysis, scene_prefab_health, build_platform_readiness, project_health, addressables, regression_trend
| Code | Meaning |
|---|---|
| 0 | No issues crossing failure threshold |
| 1 | Issues at or above fail severity |
| 2 | Invalid arguments |
| 3 | Required input missing |
| 4 | Unexpected runtime failure |
Mobile build readiness gate:
var result = UnityScannerBatch.RunBuildPlatformReadiness(new BatchOptions {
PlatformProfile = "mobile",
FailOnSeverity = "error"
});Regression gate — fail if errors increased vs baseline:
var result = UnityScannerBatch.RunRegressionTrend(new BatchOptions {
BaselinePath = "CI/baseline.json",
RegressionThreshold = 0,
FailOnSeverity = "warn"
});Addressables bundle gate:
var result = UnityScannerBatch.RunAddressables(new BatchOptions {
BuildLayoutPath = "BuildReports/BuildLayout.txt",
FailOnSeverity = "warn"
});UnityScanner includes an MCP server that lets AI coding assistants run project health scans directly from the terminal or IDE.
What it enables:
- AI-assisted project health — ask "check my project for shader issues" and get a structured report
- Automated review of non-code assets — scan before/after a change and detect regressions
- CI via AI agents — AI interprets scan results and files actionable reports
- Headless operation — runs via Unity batch mode, works on CI servers
cd path/to/MCP-Server~
npm install
npm run buildThen create a config file (see mcp-config.example.json):
{
"unityPath": "/Applications/Unity/Hub/Editor/6000.0.42f1/Unity.app/Contents/MacOS/Unity",
"projectPath": "/path/to/your/UnityProject",
"timeouts": {
"default": 120,
"runAll": 300
}
}Save as mcp-config.json in the server directory, your project root, or ~/.unityscanner/.
Claude Desktop — add to claude_desktop_config.json:
{
"mcpServers": {
"unityscanner": {
"command": "node",
"args": ["path/to/MCP-Server~/src/index.ts"]
}
}
}Cursor / OpenCode — add to your MCP settings:
{
"mcpServers": {
"unityscanner": {
"command": "node",
"args": ["./path/to/MCP-Server~/src/index.ts"]
}
}| Tool | Description |
|---|---|
unity_scanner_run_all |
Run all enabled categories — full project scan |
unity_scanner_run_category |
Run a specific category (e.g. project_health, shader_analysis) |
unity_scanner_run_selected |
Run multiple selected categories |
unity_scanner_list_categories |
List all available categories and IDs |
unity_scanner_get_settings |
Get current settings and platform profile |
unity_scanner_set_profile |
Set platform profile (mobile, console, desktop) |
unity_scanner_regression_check |
Compare scan results against a saved baseline |
unity_scanner_baseline_create |
Create a baseline snapshot for regression tracking |
"Run a full project scan and tell me what's wrong"
"Check my project for shader issues"
"Set the platform profile to mobile, then run project_health and build_platform_readiness"
"List all available scanner categories"
"Create a baseline snapshot of current project health"
The MCP server uses a bridge architecture:
AI Client → stdio JSON-RPC → Node.js MCP Server → Unity batch mode → UnityScanner API → JSON result file
Each tool call launches a fresh Unity batch mode instance (~10-30s startup). Results are written to a temp file and returned as both human-readable text and structured JSON.
Assets/Editor/UnityScanner/
Core/ — Category registry, issue model, results, settings
Categories/ — 14 category modules (scanner, mapper, settings, tab drawer)
UI/ — Main window, controls
Batch/ — Batch API entry points
MCP/ — MCP server for AI integration (C# entry points + Node.js server)
Windows/ — Standalone windows (Find References)
ContextMenus/ — [US] menu entries
Each category follows the same structure:
*Category.cs—IUnityScannerCategoryimplementation*Scanner.cs— Core scanning logic*Settings.cs— Category-specific settings/thresholds*IssueMapper.cs— Maps findings toUnityScannerIssue*TabDrawer.cs—IUnityScannerTabDrawerUI implementation*FixProvider.cs— Optional fix actions
- Using Unity's Package Manager via link https://github.com/AlexeyPerov/Unity-Scanner.git
- You can also just copy and paste all files inside Editor folder
Feel free to report bugs, request new features or to contribute to this project!
- To find unreferenced assets in Unity project see Dependencies-Hunter.
- To analyze addressables layout Addressables-Inspector.
- To find missing or empty references in your assets see Missing-References-Hunter.
- To analyze your materials and renderers see Materials-Hunter.
- To analyze your textures and atlases see Textures-Hunter.
- To analyze asset dependencies Asset-Inspector.
- Unity Editor Coroutines alternative version Lite-Editor-Coroutines.
- Simplified and compact version Pocket-Editor-Coroutines.
