Token导航 LogoToken导航

Agent Skill · 研究检索

unity-data-driven(Unity 数据 driven)

用于辅助数据整理、表格处理、CSV/Excel 分析、指标计算和图表准备。它适合让 Agent 清洗字段、汇总数据、发现异常、生成统计口径或把分析结果转成可读说明。使用时需要确认数据来源、字段含义和时间范围,避免把样本数据当全量事实;涉及敏感数据、导出文件或批量写回时,应先确认权限和脱敏边界。

unity-data-driven 是研究检索类 Agent Skill,可用宿主为 Codex、Claude、Cursor、Gemini CLI,通过GitHub 仓库获取,权限为需要联网;许可证信息还不明确,第三方扫描有提醒项,按安装量在研究检索类的 73,675 个 Skill 里排第 62,819。

总安装 326周安装 14GitHub Stars 14下载 114核验 2026-05-01

安装说明

本站只整理中文说明,不托管安装包,也不代用户安装。
GitHub
许可证
unknown

安装方式

通过对话安装

复制提示词发给支持本地命令或 Skills 的 AI 助手,先确认命令和权限,再让它执行。

请帮我安装这个 Agent Skill:unity-data-driven(Unity 数据 driven)
来源仓库:https://github.com/nice-wolf-studio/unity-claude-skills
仓库路径:skills/unity-data-driven
安装命令:
npx skills add https://github.com/nice-wolf-studio/unity-claude-skills --skill unity-data-driven
安装前请先检查当前环境是否支持对应 CLI,并向我确认将要执行的命令、安装目录、联网范围和文件读写权限;确认后再执行。

命令行安装

复制命令到本机终端执行。该命令会通过 npx skills 从第三方来源获取 Skill;本站只展示命令,不托管安装包,也不自动执行。

skills.shnpx skills
npx skills add https://github.com/nice-wolf-studio/unity-claude-skills --skill unity-data-driven

这个 Skill 是做什么的

研究检索类 Skill 主要解决「找资料、核对事实、把来源整理清楚」的问题,通常会让助手先检索再总结。

用于辅助 Unity 数据驱动开发,支持数据整理、表格处理、CSV/Excel 分析、指标计算和图表准备。适合让 Agent 清洗字段、汇总数据、发现异常或生成统计说明。使用时需确认数据来源和时间范围,避免将样本数据当作全量事实。涉及敏感数据或批量写回时,应先确认权限和脱敏边界。

unity-data-driven 属于研究检索类 Skill,可作为该场景下的辅助能力补充。

来源仓库内路径:skills/unity-data-driven

上手步骤

4 步
  1. 1

    确认自己的宿主工具

    公开资料标注支持 Codex、Claude、Cursor、Gemini CLI;先确认自己用的是其中之一,再按对应工具的 Skills 目录放置。

  2. 2

    拿到安装命令

    可以直接复制的命令是 npx skills add https://github.com/nice-wolf-studio/unity-claude-skills --skill u…;命令里的占位符换成自己的路径或密钥再执行。

  3. 3

    安装后重启宿主

    命令执行完要完全退出宿主工具再打开,让新装的 Skill 被重新扫描到,否则列表里可能看不到。

  4. 4

    按最小权限试用

    它需要需要联网,先在测试目录或测试账号里跑,确认没问题再放到正式项目。

常见问题

6 条

unity-data-driven 怎么安装?

在终端执行 npx skills add https://github.com/nice-wolf-studio/unity-claude-skills --skill unity-data-…。装完完全退出宿主工具再打开,让它重新扫描本地 Skill。

unity-data-driven 支持哪些 AI 工具?

公开资料标注支持 Codex、Claude、Cursor、Gemini CLI。其他工具只要能读取本地 Skill 目录,一般也能手动放进去,但要自己确认触发方式是否一致。

unity-data-driven 需要什么权限?

权限级别是需要联网。该 Skill 可能需要联网访问来源站点、仓库或外部 API;具体网络访问范围需要结合源码和 README 复核。

unity-data-driven 是免费的吗?

公开资料没有标注明确许可证,收费和商用限制都无法确认。正式使用前请到来源仓库核对授权说明。

unity-data-driven 用的人多吗?

公开数据是累计安装 326 次,近一周安装 14 次,下载 114 次,GitHub 14 星。这些数字来自第三方目录统计,只能当作热度参考,不代表质量或安全性。

unity-data-driven 安全吗?

第三方扫描有提醒项,说明存在需要人工确认的行为。安装前请打开源码看清它读了什么、联网到哪里。

值得选的地方和要先想清楚的地方

值得选的地方

  • 同时适配 4 种工具,换客户端不用重新找替代品。

先想清楚的地方

  • 许可证没有明确标注,能不能商用、能不能再分发都不确定。
  • 第三方扫描存在提醒项或未通过项,需要人工逐条确认。
  • 第三方目录整理,版本更新和行为细节以来源仓库为准。

适不适合你

更适合

  • 已经在用 Codex,想直接扩展它能力的人。
  • 需要处理研究检索类任务、又不想每次都手写提示词的人。

不太适合

  • 需要明确授权才能用到公司项目里的场景。
  • 需要厂商级技术支持和 SLA 的生产系统。

研究检索类 Skill 对比

6 个同类
Skill总安装GitHub 星权限许可证扫描
find-skills

寻找技能 · Codex、Claude

31,199,81316,486操作浏览器MIT有提醒
self-improving-agent

自我改进 Agent · OpenClaw

16,464,0023,396接触敏感数据MIT-0通过
soultrace

灵魂追踪 · Codex、Claude

7,588,94412需要联网未明确有提醒
self-improving

自我完善 · OpenClaw

6,829,2841,046需要联网MIT-0通过
microsoft-foundry

微软代工厂 · Codex、Claude

6,614,144764调用外部服务未明确有提醒
unity-data-driven

Unity 数据 driven · 本页

32614需要联网未明确有提醒

按安装量取同分类前几名;数字来自第三方目录统计,只作热度参考,不代表质量结论。

SKILL.md 原文

Data-Driven Design -- Decision Patterns

Prerequisite skills: unity-scripting/references/scriptableobjects.md (SO creation, event channels, runtime sets, variable references), unity-editor-tools (custom inspectors, EditorWindow)

These patterns address the most common data management failure: Claude hardcodes tunable values in MonoBehaviours or uses magic numbers, making iteration and balancing painful. All game data should be designer-editable without code changes.


PATTERN: Config Data Storage Selection

WHEN: Choosing where to store game configuration (enemy stats, level layouts, loot tables)

DECISION:

  • ScriptableObject assets (default) -- Designer-editable in Inspector, type-safe, refactorable, version-controlled as YAML. Best for most game configs.
  • JSON/CSV files -- External tools generate data (spreadsheets, web tools), need modding support, or bulk editing is required. Loaded at runtime.
  • Embedded constants -- Truly fixed values (physics constants, math, protocol versions). static readonly or const.

SCAFFOLD (ScriptableObject config):

[CreateAssetMenu(fileName = "New Enemy Config", menuName = "Game/Enemy Config")]
public class EnemyConfig : ScriptableObject
{
    [Header("Stats")]
    [Min(1)] public int maxHealth = 100;
    [Range(0f, 20f)] public float moveSpeed = 5f;
    [Range(0f, 50f)] public float attackDamage = 10f;

    [Header("AI")]
    [Range(1f, 50f)] public float detectionRange = 15f;
    [Range(0.5f, 5f)] public float attackCooldown = 1.5f;

    [Header("Rewards")]
    [Min(0)] public int xpReward = 25;
    [Tooltip("Loot dropped on death. Leave empty for no loot.")]
    public LootTable lootTable;
}

// Usage in MonoBehaviour:
public class Enemy : MonoBehaviour
{
    [SerializeField] private EnemyConfig config; // Assign in Inspector

    void Awake()
    {
        // Read from config -- never hardcode values
        _health = new HealthSystem(config.maxHealth);
        _moveSpeed = config.moveSpeed;
    }
}

GOTCHA: ScriptableObject assets modified at runtime in the Editor persist those changes to disk (the.asset file is saved). In builds, runtime modifications are lost when the application exits. Never rely on runtime SO modification for save data -- use a separate save system. See unity-save-system for persistence.


PATTERN: ScriptableObject Inheritance vs Composition

WHEN: Multiple config types share base fields (all enemies have HP/speed, subtypes add flying/ranged/boss fields)

DECISION:

  • SO inheritance -- EnemyConfig: ScriptableObject, FlyingEnemyConfig: EnemyConfig. Each subtype adds its own fields. Designer sees only relevant fields per asset. Works well for 2-3 levels of hierarchy.
  • Nested [Serializable] structs (composition) -- EnemyConfig contains [SerializeField] MovementConfig movement; [SerializeField] AttackConfig attack;. Mix-and-match capabilities. Often cleaner than inheritance for wide variation.

SCAFFOLD (Composition):

[System.Serializable]
public struct MovementConfig
{
    [Range(0f, 20f)] public float speed;
    public bool canFly;
    [Tooltip("Only used if canFly is true")]
    [Range(0f, 50f)] public float flyHeight;
}

[System.Serializable]
public struct AttackConfig
{
    public AttackType type;
    [Range(0f, 100f)] public float damage;
    [Range(0.1f, 10f)] public float cooldown;
    [Range(0f, 30f)] public float range;
}

[CreateAssetMenu(fileName = "New Enemy", menuName = "Game/Enemy Config")]
public class EnemyConfig : ScriptableObject
{
    [Header("Identity")]
    public string displayName;
    [TextArea(2, 4)] public string description;

    [Header("Movement")]
    public MovementConfig movement;

    [Header("Combat")]
    public AttackConfig attack;

    [Header("Stats")]
    [Min(1)] public int maxHealth = 100;
    [Min(0)] public int xpReward = 25;
}

SCAFFOLD (Inheritance):

public abstract class EnemyConfig : ScriptableObject
{
    [Min(1)] public int maxHealth = 100;
    [Range(0f, 20f)] public float moveSpeed = 5f;
}

[CreateAssetMenu(fileName = "New Melee Enemy", menuName = "Game/Enemies/Melee")]
public class MeleeEnemyConfig : EnemyConfig
{
    [Range(0f, 50f)] public float meleeDamage = 15f;
    [Range(0.5f, 3f)] public float attackRadius = 1.5f;
}

[CreateAssetMenu(fileName = "New Ranged Enemy", menuName = "Game/Enemies/Ranged")]
public class RangedEnemyConfig : EnemyConfig
{
    [Range(0f, 50f)] public float projectileDamage = 10f;
    [Range(5f, 30f)] public float fireRange = 15f;
    public GameObject projectilePrefab;
}

GOTCHA: SO inheritance shows ALL base fields in the Inspector (no hiding). Use [HideInInspector] or a custom editor if base fields clutter the Inspector. Composition with [Serializable] structs is often cleaner because each struct can be reused across unrelated config types. [CreateAssetMenu] only works on concrete (non-abstract) ScriptableObject subclasses.


PATTERN: JSON Data Pipeline

WHEN: Loading game data from external sources (spreadsheets, web APIs, modding)

DECISION:

  • Build-time pipeline -- Convert JSON/CSV to ScriptableObject assets in an Editor script. Data is baked at build time. Designer edits spreadsheet, runs import, data becomes SO assets.
  • Runtime loading -- Load JSON from StreamingAssets or a remote URL at runtime. Supports modding, hot-reload, and server-driven config.

SCAFFOLD (Runtime JSON loading):

// Data class (plain C#, not ScriptableObject)
[System.Serializable]
public class WaveData
{
    public int waveNumber;
    public string[] enemyTypes;
    public int[] enemyCounts;
    public float spawnInterval;
}

[System.Serializable]
public class WaveDataList
{
    public WaveData[] waves; // JsonUtility requires a wrapper for arrays
}

// Loading
public class WaveLoader : MonoBehaviour
{
    async Awaitable<WaveData[]> LoadWaves()
    {
        string path = Path.Combine(Application.streamingAssetsPath, "waves.json");

        #if UNITY_ANDROID && !UNITY_EDITOR
        // Android: StreamingAssets is inside APK, must use UnityWebRequest
        using var request = UnityWebRequest.Get(path);
        await request.SendWebRequest();
        string json = request.downloadHandler.text;
        #else
        string json = await File.ReadAllTextAsync(path);
        #endif

        var wrapper = JsonUtility.FromJson<WaveDataList>(json);
        return wrapper.waves;
    }
}

GOTCHA: JsonUtility cannot serialize: dictionaries, top-level arrays (need a wrapper class), polymorphic types, null values (uses default instead), or properties (only fields). For any of these, use Newtonsoft JSON (com.unity.nuget.newtonsoft-json package). Application.streamingAssetsPath is read-only on all platforms and requires UnityWebRequest on Android. See references/data-pipeline-patterns.md for the Editor import script.


PATTERN: Designer Handoff Workflow

WHEN: Designers need to create and edit game data without touching code

DECISION:

  • SO + CreateAssetMenu -- Designers right-click in Project window to create config assets, edit in Inspector. Best for small-medium data sets (< 100 items).
  • SO + Custom EditorWindow -- Bulk editing, validation, search/filter. Worth the investment for large data sets (100+ items, loot tables, dialogue).
  • External spreadsheet + import -- Google Sheets/Excel -> CSV -> Editor import script. Best when designers prefer spreadsheets or data comes from external tools.

SCAFFOLD (Self-documenting Inspector):

[CreateAssetMenu(fileName = "New Weapon", menuName = "Game/Weapons/Weapon Config")]
public class WeaponConfig : ScriptableObject
{
    [Header("Identity")]
    [Tooltip("Display name shown in UI and inventory")]
    public string displayName;

    [TextArea(2, 5)]
    [Tooltip("Flavor text shown in the item tooltip")]
    public string description;

    public Sprite icon;

    [Header("Combat Stats")]
    [Tooltip("Base damage before modifiers. Actual damage = base * level multiplier")]
    [Range(1f, 200f)] public float baseDamage = 10f;

    [Tooltip("Seconds between attacks")]
    [Range(0.1f, 5f)] public float attackSpeed = 1f;

    [Tooltip("Maximum range in world units")]
    [Range(0.5f, 30f)] public float range = 2f;

    [Header("VFX/SFX")]
    [Tooltip("Played on hit. Leave null for no effect")]
    public GameObject hitEffect;

    [Tooltip("Played on attack")]
    public AudioClip attackSound;

    [Space(10)]
    [Header("Advanced")]
    [Tooltip("Damage falloff over distance. X = normalized distance (0-1), Y = damage multiplier")]
    public AnimationCurve damageFalloff = AnimationCurve.Linear(0, 1, 1, 0.5f);
}

GOTCHA: Designers cannot read code comments -- use [Tooltip("...")] on every field. Use [Range(min, max)] to prevent invalid data entry. Use [Header("Section")] to group related fields. Use [Space(10)] for visual separation. AnimationCurve fields are powerful for designer-tunable falloffs, easing, and response curves. Mark optional references with tooltips like "Leave null for no effect".


PATTERN: Data Versioning and Migration

WHEN: Shipped data format changes between updates (added/renamed/removed fields)

DECISION:

  • SO resilient serialization (additive changes) -- Adding new fields gives them default values. Removing fields silently drops data. Safe for non-breaking changes. Unity handles this automatically.
  • Explicit version field + migrator -- For breaking changes (renamed fields, restructured data). Store int version in the data, run migration on load.

SCAFFOLD (FormerlySerializedAs for renames):

using UnityEngine.Serialization;

public class EnemyConfig : ScriptableObject
{
    // Renamed from "hp" to "maxHealth" -- existing assets preserved
    [FormerlySerializedAs("hp")]
    [Min(1)] public int maxHealth = 100;

    // Renamed from "speed" to "moveSpeed"
    [FormerlySerializedAs("speed")]
    [Range(0f, 20f)] public float moveSpeed = 5f;

    // New field -- existing assets get the default value (25)
    [Min(0)] public int xpReward = 25;

    // Removed field: just delete it. Existing assets silently drop the old data.
}

SCAFFOLD (Versioned runtime data with migration):

[System.Serializable]
public class PlayerData
{
    public int version = 2; // Current schema version
    public string playerName;
    public int level;
    public float[] position; // v2: changed from Vector3 to float[] for JSON compat

    // Migration from v1 to v2
    public static PlayerData MigrateFromV1(string json)
    {
        // v1 had "pos_x", "pos_y", "pos_z" as separate fields
        var jObj = Newtonsoft.Json.Linq.JObject.Parse(json);
        if ((int)jObj["version"] == 1)
        {
            float x = (float)jObj["pos_x"];
            float y = (float)jObj["pos_y"];
            float z = (float)jObj["pos_z"];
            jObj.Remove("pos_x"); jObj.Remove("pos_y"); jObj.Remove("pos_z");
            jObj["position"] = new Newtonsoft.Json.Linq.JArray(x, y, z);
            jObj["version"] = 2;
        }
        return jObj.ToObject<PlayerData>();
    }
}

GOTCHA: [FormerlySerializedAs] only works for Unity serialization (Inspector,.asset files, prefabs). It does NOT work for JSON or custom serialization. For JSON migration, you must parse the raw JSON before deserializing to the new type. SO data versioning and save data versioning are different problems -- see unity-save-system for save file migration.


Inspector Attribute Quick Reference

AttributePurposeExample
[Header("Section")]Section labelGroup related fields
[Tooltip("Help text")]Hover help textExplain non-obvious fields
[Range(min, max)]Slider for numeric fields[Range(0, 100)]
[Min(value)]Minimum value clamp[Min(0)] for non-negative
[TextArea(min, max)]Multi-line text field[TextArea(2, 5)]
[Space(pixels)]Vertical spacing[Space(10)]
[HideInInspector]Hide from InspectorInternal fields
[FormerlySerializedAs]Preserve data on renameMigration safety
[ColorUsage(alpha, hdr)]Color picker config[ColorUsage(true, true)]
[GradientUsage(hdr)]Gradient HDR supportVFX color ramps
[Delayed]Apply on Enter/focus lossPrevent per-keystroke updates
[SerializeReference]Polymorphic serializationInterface fields

Related Skills

  • unity-scripting/references/scriptableobjects.md -- SO creation, event channels, runtime sets, variable references
  • unity-editor-tools -- Custom EditorWindow for bulk editing, PropertyDrawer for custom display
  • unity-save-system -- Runtime data persistence (versus config data)
  • unity-game-architecture -- ScriptableObject architecture decisions

Additional Resources

平台分布

Codex

35.68%
按下载量换算41

Claude

30.67%
按下载量换算35

Cursor

18.34%
按下载量换算21

Gemini CLI

10.71%
按下载量换算12

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

可疑

权限和风险

需要联网

该 Skill 可能需要联网访问来源站点、仓库或外部 API;具体网络访问范围需要结合源码和 README 复核。

安装前确认

本站仅展示第三方公开信息,不托管安装包,不提供自动安装或运行环境。安装前应自行审查源码、依赖和命令行为。来源安全扫描存在提醒或未通过项,安装前请自行复核。目前只有一个公开来源,建议同时核对源仓库。

继续浏览同类 Skills