plugins/unity-master/skills/unity-modding/SKILL.md
Unity game modding architecture and mod-loading systems. PROACTIVELY activate for: (1) adding modding support to a Unity game, (2) Asset Bundles for mods, (3) Addressables for mod content, (4) Lua scripting via MoonSharp, (5) Harmony patching for runtime IL injection, (6) Steam Workshop integration, (7) mod manager UI, (8) plugin architecture for user-generated content (UGC), (9) custom content loading at runtime, (10) sandboxing and security for untrusted mods. Provides: AssetBundle vs Addressables tradeoffs, MoonSharp setup, Harmony patch patterns, Steam Workshop SDK integration, mod manager templates, and security guidelines.
npx skillsauth add JosiahSiegel/claude-plugin-marketplace unity-moddingInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
Guide for designing Unity games that support community modding. Covers asset loading systems, scripted mod APIs, mod manager patterns, distribution platforms, and techniques for modding existing games.
| Tier | What Modders Can Do | Complexity | |------|---------------------|------------| | Data Mods | Replace textures, sounds, configs (JSON/XML) | Low | | Asset Mods | Add new models, maps, items via Asset Bundles | Medium | | Script Mods | Custom game logic via Lua/scripting API | High | | Code Mods | Patch game assemblies via Harmony/BepInEx | Expert |
Design your game to support at least Tier 1-2 from the start. Tiers 3-4 require deliberate API design.
Addressables provide a unified API for loading assets from local bundles, remote servers, or mod directories.
// Load a mod's asset catalog at runtime
public async Awaitable LoadModCatalog(string modPath)
{
string catalogPath = Path.Combine(modPath, "catalog.json");
var locator = await Addressables.LoadContentCatalogAsync(catalogPath);
Debug.Log($"Loaded mod catalog with {locator.Keys.Count()} assets");
}
// Load an asset by address (works for both base game and mods)
var prefab = await Addressables.LoadAssetAsync<GameObject>("enemies/goblin");
Instantiate(prefab);
Mod workflow:
public async Awaitable<AssetBundle> LoadModBundle(string path)
{
var request = AssetBundle.LoadFromFileAsync(path);
await request;
return request.assetBundle;
}
// Load specific asset from bundle
var bundle = await LoadModBundle("mods/weapons/swords.bundle");
var swordPrefab = bundle.LoadAsset<GameObject>("BroadSword");
Key rules:
bundle.Unload(true)AssetBundle.LoadFromFileAsync for local files (not LoadFromMemory)For simple data mods (JSON configs, CSV tables):
public T LoadModConfig<T>(string modPath, string fileName)
{
string filePath = Path.Combine(modPath, fileName);
if (!File.Exists(filePath)) return default;
string json = File.ReadAllText(filePath);
return JsonUtility.FromJson<T>(json);
}
Use FileSystemWatcher to detect changes and hot-reload during development.
MoonSharp is a Lua interpreter written in C# that runs on all Unity platforms.
Install via NuGet or download the DLL. Register safe API surfaces for modders:
public class ModScriptEngine
{
Script _lua;
public void Initialize()
{
_lua = new Script();
// Register safe game API
_lua.Globals["SpawnEntity"] = (Func<string, float, float, float, bool>)SpawnEntity;
_lua.Globals["GetPlayerHealth"] = (Func<float>)(() => player.Health);
_lua.Globals["ShowMessage"] = (Action<string>)ShowMessage;
_lua.Globals["RegisterEvent"] = (Action<string, DynValue>)RegisterEvent;
}
public void LoadMod(string luaFilePath)
{
string code = File.ReadAllText(luaFilePath);
_lua.DoString(code);
}
bool SpawnEntity(string id, float x, float y, float z)
{
// Validate and spawn - never trust mod input
if (!entityRegistry.Contains(id)) return false;
var pos = new Vector3(x, y, z);
if (!IsValidSpawnPosition(pos)) return false;
entityRegistry.Spawn(id, pos);
return true;
}
}
-- my_mod/init.lua
RegisterEvent("OnPlayerEnterZone", function(zoneName)
if zoneName == "boss_arena" then
SpawnEntity("dragon_boss", 0, 5, 0)
ShowMessage("A dragon appears!")
end
end)
| Risk | Mitigation |
|------|-----------|
| File system access | Do NOT expose System.IO to Lua |
| Infinite loops | Set instruction count limits: script.Options.InstructionLimit = 100000 |
| Memory abuse | Limit table sizes, monitor allocation |
| Network access | Never expose HTTP/socket APIs |
| Reflection abuse | MoonSharp sandboxes by default, but validate registered types |
Game
├── ModManager (singleton)
│ ├── DiscoverMods(modsFolder) // Scan for mod manifests
│ ├── ValidateMod(manifest) // Version compat, dependencies
│ ├── LoadMod(modId) // Load assets + scripts
│ ├── UnloadMod(modId) // Clean teardown
│ └── GetLoadOrder() // Dependency-sorted order
├── ModManifest (ScriptableObject / JSON)
│ ├── id, name, version, author
│ ├── dependencies[]
│ ├── gameVersionMin / gameVersionMax
│ └── entryPoint (script path)
└── ModSandbox
├── Lua scripting environment
└── Restricted API surface
{
"id": "com.author.mymod",
"name": "My Awesome Mod",
"version": "1.2.0",
"author": "ModAuthor",
"description": "Adds new enemies and weapons",
"gameVersionMin": "1.0.0",
"gameVersionMax": "2.0.0",
"dependencies": ["[email protected]"],
"entryPoint": "scripts/init.lua",
"assets": "bundles/"
}
Use Steamworks.NET or Facepunch.Steamworks for Steam Workshop integration:
SteamUGC.CreateItem + SteamUGC.SubmitItemUpdateSteamUGC.GetSubscribedItems() at startupSteamUGC.GetItemInstallInfo path into ModManagerHarmony is used by modders (via BepInEx/MelonLoader) to patch compiled game code at runtime.
[HarmonyPatch(typeof(PlayerHealth), "TakeDamage")]
class DamageModifier
{
// Prefix: runs before original method
static bool Prefix(ref float damage)
{
damage *= 0.5f; // Halve all damage
return true; // true = continue to original
}
// Postfix: runs after original method
static void Postfix(PlayerHealth __instance)
{
Debug.Log($"Health after damage: {__instance.CurrentHealth}");
}
}
If designing a game intended to be modded via Harmony, avoid aggressive code obfuscation and keep method signatures stable across updates.
references/mod-framework-detail.md -- Complete mod manager implementation, dependency resolution algorithm, mod load ordering, versioned API patterns, hot-reload systems, mod testing frameworks, BepInEx/MelonLoader setup guidedevelopment
This skill should be used when the user asks to train, debug, scale, or improve ML models. PROACTIVELY activate for: (1) PyTorch, TensorFlow/Keras, JAX, Flax, Hugging Face Trainer/Accelerate training loops, (2) distributed training, DDP/FSDP/DeepSpeed, TPU/GPU setup, (3) mixed precision AMP/bf16, gradient accumulation, checkpointing, seeding, (4) overfitting, imbalance, loss functions, regularization, LR schedules, warmup, (5) memory optimization, gradient checkpointing, offloading, quantization-aware training. Provides: reproducible training best practices across deep learning and classical ML.
development
This skill should be used when the user asks to productionize, track, version, govern, monitor, or automate ML systems. PROACTIVELY activate for: (1) MLflow, Weights & Biases, Neptune, Comet, ClearML experiment tracking, (2) model registry, model versioning, artifact lineage, reproducibility, (3) Kubeflow, SageMaker Pipelines, Vertex AI Pipelines, Azure ML pipelines, Databricks workflows, (4) CI/CD, continuous training/evaluation, A/B tests, canary/shadow deployments, (5) drift detection, model monitoring, data validation, responsible AI governance. Provides: end-to-end MLOps architecture and operational safeguards.
development
This skill should be used when the user asks to optimize, export, serve, compress, or accelerate ML inference. PROACTIVELY activate for: (1) latency, throughput, p95/p99, batching, concurrency, KV cache, memory, or cost issues, (2) quantization INT8/INT4, GPTQ, AWQ, bitsandbytes, pruning, sparsity, distillation, (3) ONNX export, ONNX Runtime, TensorRT, TorchScript, torch.compile, XLA, OpenVINO, Core ML, TFLite, (4) Triton, TorchServe, TF Serving, BentoML, Seldon, KServe configuration, (5) edge deployment, CPU/GPU/TPU/Inferentia serving. Provides: hardware-aware inference optimization and safe benchmarking.
testing
This skill should be used when the user asks to tune hyperparameters, run sweeps, optimize search spaces, or use AutoML. PROACTIVELY activate for: (1) Optuna, Ray Tune, FLAML, AutoGluon, Hyperopt, Nevergrad, KerasTuner, W&B sweeps, (2) grid search, random search, Bayesian optimization, TPE, Gaussian processes, evolutionary search, (3) ASHA, Hyperband, successive halving, multi-fidelity optimization, population-based training, (4) learning-rate finder, batch-size search, early stopping, pruning, (5) reproducible sweep design and experiment analysis. Provides: budget-aware hyperparameter search strategy.