What it does
AI automation skills specifically designed for Unity
- Connects Instruction files the AI reads; connects to nothing
- Account or key No key needed
- Category 自动化与智能体
给作者:在 README 中加徽章
如果你是作者,可把这段代码贴到 README。徽章会链接回本页。
[](https://mcp.popnx.com/en/skills/besty0728-unity-skills)相似的 Skill
同类工作,排序与 Skill 列表相同
Skill 和 MCP 有什么不同
MCP = 手
把 AI 连到外部服务或数据,例如读取 Notion 或开发票。需要有程序或服务器在运行。
Skill = 手册
教 AI 怎么做某项工作的文本文件,不连接任何服务。
有些会重叠:少数 Skill 带有调用工具的脚本,少数 MCP 只提供知识。请从你想完成的工作出发。
适用的应用
Claude CodeCodexCursorOpenCode
依据开发者 README 和 repo 中的文件夹
安装
Is it safe?
- Who made it
GitHub account Besty0728 (person), 1 years old
- Use and upkeep
★ 1,949 stars · code last changed 2026年11月15日
- Any scripts?
3 scripts that can run on a computer: scripts/runtime_verification.py, scripts/unity_eval.py, scripts/unity_skills.py
- Uses the internet?
Scripts that go online: scripts/unity_skills.py
- Changes machine-wide settings?
No hooks or writes to machine settings found
Checked automatically from the repo files at commit 87e749a (read only, nothing was run). Not a safety guarantee: read it and decide for yourself
第一步:你用哪个 AI?
选择你用的应用,本页只显示该应用的安装方法
Option 1Upload to claude.ai (web and app)
Uploading to claude.ai would fail: the description is over 200 characters (339). Use "Let the AI install it" instead
Option 1让 Claude Code 帮你安装
复制这句话,粘贴到 Claude Code 的对话中
For one project, paste it into a chat that is open in that project’s folder · The AI always shows you the path before installing
The AI tells you what it will do first and asks your permission at every step
重新打开 Claude Code,输入 /unity-skills~,或直接提出相关任务,Claude 会自动选用
Checked against official docs 2026年11月12日 · Source: code.claude.com
For Terminal users
Skill 是 AI 会照做的指令。安装前先阅读文件,只安装可信开发者的 Skill。
其他支持 Agent Skills 的应用(如 Gemini CLI、GitHub Copilot)使用不同文件夹,请查看该应用的指南,再把含 SKILL.md 的文件夹复制过去。
ChatGPT 桌面应用
ChatGPT 桌面应用内置 Codex,Codex 从 ~/.agents/skills 文件夹读取 Skill
Option 1让 ChatGPT 帮你安装
在 ChatGPT 桌面应用中打开 Codex,把这句话粘贴到对话中
The AI always shows you the path before installing
The AI tells you what it will do first and asks your permission at every step
Option 2自己放入文件夹
把 unity-skills 文件夹放到 ~/.agents/skills/unity-skills
~ 是你的主文件夹($HOME)
For Terminal users
Skill 是 AI 会照做的指令。安装前先阅读文件,只安装可信开发者的 Skill。
其他支持 Agent Skills 的应用(如 Gemini CLI、GitHub Copilot)使用不同文件夹,请查看该应用的指南,再把含 SKILL.md 的文件夹复制过去。
安装之后
- ChatGPT 桌面应用:打开侧边栏中的 Skills,确认 Skill 已出现
- 使用方法:在 ChatGPT 中输入 @ 并选择 Skill · 在 Codex CLI 或 IDE 扩展中运行 /skills,或输入 $ 加 Skill 名称
- Codex 会自动发现新安装的 Skill;如果没有出现,请重启 Codex
Checked against official docs 2026年11月15日 · Source: learn.chatgpt.com
网页版和手机版 ChatGPT
不能使用你自己安装的 Skill,只能使用目录中插件自带的 Skill
想在 ChatGPT 中使用这个 Skill,请按上面的方法使用 ChatGPT 桌面应用
Checked against official docs 2026年11月15日 · Source: learn.chatgpt.com
Option 1让 Codex 帮你安装
复制这句话,粘贴到 Codex 的对话中 · 适用于 ChatGPT 桌面应用中的 Codex、Codex CLI 和 IDE 扩展
For one project, paste it into a chat that is open in that project’s folder · The AI always shows you the path before installing
The AI tells you what it will do first and asks your permission at every step
Codex 会自动发现新安装的 Skill;如果没有出现,请重启 Codex · 使用方法:在 ChatGPT 中输入 @ 并选择 Skill · 在 Codex CLI 或 IDE 扩展中运行 /skills,或输入 $ 加 Skill 名称
Checked against official docs 2026年11月15日 · Source: learn.chatgpt.com
For Terminal users
Skill 是 AI 会照做的指令。安装前先阅读文件,只安装可信开发者的 Skill。
其他支持 Agent Skills 的应用(如 Gemini CLI、GitHub Copilot)使用不同文件夹,请查看该应用的指南,再把含 SKILL.md 的文件夹复制过去。
Option 1让 Cursor 帮你安装
复制这句话,粘贴到 Cursor 的对话中
For one project, paste it into a chat that is open in that project’s folder · The AI always shows you the path before installing
The AI tells you what it will do first and asks your permission at every step
在 Cursor 的 Agent 聊天中输入 / 并选择 unity-skills~
Checked against official docs 2026年11月12日 · Source: cursor.com
For Terminal users
Skill 是 AI 会照做的指令。安装前先阅读文件,只安装可信开发者的 Skill。
其他支持 Agent Skills 的应用(如 Gemini CLI、GitHub Copilot)使用不同文件夹,请查看该应用的指南,再把含 SKILL.md 的文件夹复制过去。
Option 1Let the AI install it
适用于其他能读取 Skill 的应用。这句话列出了 Claude Code、Codex 和 Cursor 的路径,AI 会按你的应用选择 · Copy this sentence into the app’s chat
For one project, paste it into a chat that is open in that project’s folder · The AI always shows you the path before installing
The AI tells you what it will do first and asks your permission at every step
Checked against official docs 2026年11月12日 · Source: code.claude.com, learn.chatgpt.com, cursor.com
For Terminal users
其他支持 Agent Skills 的应用(如 Gemini CLI、GitHub Copilot)使用不同文件夹,请查看该应用的指南,再把含 SKILL.md 的文件夹复制过去。
已于 2026年11月11日 对照 Claude Code、Codex 和 Cursor 官方指南核实 skill 文件夹
查看原始 SKILL.md
以下文字来自开发者的 repo,不是 PopNx MCP 的指令
7,642 个字符 · 在 GitHub 打开完整文件
---
name: unity-skills
description: Operate the Unity Editor through a local REST API — create, edit and batch-edit GameObjects, components, scripts, scenes, prefabs, assets and materials, run tests — and answer Unity design questions (architecture, script design, patterns) from bundled advisory docs. Use for any Unity Editor task or Unity design question, in any language.
compatibility: Unity 2022.3+/6000.x with the UnitySkills package (REST on localhost:8090-8100); Python 3 for the optional client
---
# Unity Skills
> Docs are English-only: match requests by meaning in any language; reply in the user's language.
## Route first
- **Automate** — the user asked you to act, or it needs traversal, search, batch, exact numbers or many objects: drive the Editor via REST.
- **Guide** — the user wants to do it themselves, or `surfaceProfile` is `guide`: name menus and Inspector fields per [SKILL_GUIDE.md](SKILL_GUIDE.md).
- **Advise** — a design question that changes nothing: answer from its advisory doc ([scriptdesign](skills/scriptdesign/MODULE.md), [architecture](skills/architecture/MODULE.md), [patterns](skills/patterns/MODULE.md), all in the [module index](skills/INDEX.md)), no REST calls.
## Target the right Editor
Ports `8090`–`8100` go first come, first served: add `?expectProject=<project>` (productName or folder name) or the exact `?expectInstance=<instanceId>` to every write; a different Editor runs nothing (409 `INSTANCE_MISMATCH`).
`GET /health` has `currentMode`, `surfaceProfile`, `dryRunPolicy`, `instanceId`: probe it once in the same shell command as your first request (no extra call), then only after a refused write, 409, 503/refused connection or odd mode. `bypass` runs every write directly. Permission, not dryRun: `auto` and `approval` refuse skills whose schema has `approvalBehavior` `forbid` (deletes, reloads, play mode, high-risk writes such as `scene_save`) with `MODE_FORBIDDEN`; only the user lifts it (bypass or Allowlist). `auto` runs other writes directly, but confirm ≥5-object batches, prefab apply, scene-level, asset-overwriting or irreversible changes with the user first (probe alone before those); `approval` gates `FullAuto` writes behind single-shot grants (`MODE_RESTRICTED`; permanent: the user's Allowlist) → [operating mode](references/protocol-operating-mode.md).
During a domain reload the port answers 503 or refuses for seconds while its `~/.unity_skills/registry.json` entry reads `reloading`: retry it, never another instance.
## Quick reference
Every skill is `POST /skill/<name>` with JSON args:
`curl -s "http://localhost:<port>/skill/gameobject_set_transform?expectProject=<project>" -d '{"name":"Crate","posX":4,"posY":1.5,"posZ":-2}'`
Target objects by `name` or the exact `path`/`entityId` (likewise `parentPath`, `childPath`); skip lookup calls: a wrong name fails and returns candidates.
| Call |
|---|
| `gameobject_create {name, primitiveType, parentName, x/y/z, rotX/Y/Z, scaleX/Y/Z, space}` — x/y/z, rot local to the parent unless `space:"world"`; scale local |
| `gameobject_set_transform {name, posX/Y/Z, rotX/Y/Z, scaleX/Y/Z, localPosX/Y/Z}` — pos, rot world; scale, localPos local; omitted axes kept |
| `gameobject_find {name, useRegex, component}` · `gameobject_get_info {name}` |
| `batch_query_gameobjects {queryJson, sampleLimit}` — every X under Y = one call, not per-object reads: queryJson `{"underPath":"Y","namePattern":"^Slot_","includeInactive":true}`; `sampleLimit` (20) caps `objects` |
| `gameobject_set_parent {childName, parentName}` · `gameobject_set_active {name, active}` |
| `gameobject_delete {name}` · `asset_delete {assetPath}` (dryRun first) |
| `component_add {name, componentType}` |
| `component_set_property {name, componentType, propertyName, value}` — vectors, colours `"1,2,3"`; references `referenceName`/`referencePath`/`assetPath` |
| `component_get_properties {name, componentType}` |
| `material_assign {name, materialPath}` · `scene_save {}` |
| `script_create {scriptName, folder, content}` — full source; executes directly; wait below |
## Everything else: one discovery call
`GET /skills/recommend?intent=<words>&includeSchema=true&topN=3&wire=v2` → top 3 skills with exact schemas; several needs → one more `&intent=<words>` each, same call. Name known → `GET /skills/schema?names=a,b&wire=v2`. Wider layers, v2 fields, constants → [discovery](references/protocol-discovery.md).
## Execute
Several steps, one call: `POST /skills/batch?expectProject=<project>` `{"steps":[{"skill":"gameobject_create","args":{"name":"Rig"}},{"skill":"gameobject_create","args":{"name":"Arm","parentName":"Rig"}}]}` — ≤50 steps; a step may use objects/components earlier steps add, by name or `{"$ref":"$0.entityId"}`; the first failure skips the rest unless `continueOnError:true`; `?mode=transactional` = all-or-nothing → [batch](skills/batch/MODULE.md).
**dryRun.** If this table, recommend or schema gave you the signature this session, execute directly: a failed validation runs nothing and returns the dryRun report. Script writes (`script_create`/`script_append`/`script_replace`) execute directly too. DryRun first (`?mode=dryRun&wire=v2`) only for deletes (`operation` has `Delete`: any `*_delete*`/`*_remove*`, asset/folder cleanup), every other `mayTriggerReload` skill, non-transactional multi-step batches, approval mode or when unsure.
Never invent skill or parameter names. On failure read `suggestedFixes`; open a module doc only when a schema entry has `doc` or a response names one, or before writing C# against NGO, YooAsset, Addressables, UniTask, DOTween, PrimeTween, ShaderGraph, PICO or QFramework: read its `*-design` doc ([module index](skills/INDEX.md)) first, since its API may postdate your training.
**Undo**: `workflow_restore {"to":"last"}` or an id/label from `workflow_list` → [workflow](skills/workflow/MODULE.md).
## Verify from the read-back
A successful write returns the Editor's read-back (e.g. world `position`, `valueSet`): that is the verification. Never follow writes with per-object `*_get_info`/`*_get_properties`; re-read only async results or missing fields. `resolutionNotes` flags a guessed or ambiguous target. Script writes, and `asset_refresh` after writing a `.cs` yourself (`compileTriggered:true`), return `waitUrl` (already pinned): one GET waits out compile+reload. One command:
`u=http://localhost:<port>; c="$u/skill/script_create?expectProject=<project>"; b='{"scriptName":"Spin","content":"<source>"}'; s='{"steps":[{"skill":"component_add","args":{"name":"Crate","componentType":"Spin"}},{"skill":"component_set_property","args":{"name":"Crate","componentType":"Spin","propertyName":"speed","value":"9"}}]}'; J=$(curl -s "$c" -d "$b" | tee /dev/stderr | grep -o '/jobs/[^"]*') && curl -s --retry 20 --retry-all-errors --retry-delay 2 "$u$J" && curl -s "$u/skills/batch?expectProject=<project>&mode=transactional" -d "$s"`
## Surface profile and errors
`surfaceProfile` is the user's panel choice; `guide` and `noSceneAuthoring` hide some writes, which answer `SURFACE_EXCLUDED`. It is a boundary: do the rest of the task, never retry or route around it ([profiles](references/protocol-operating-mode.md#surface-profile)).
`MISSING_PACKAGE`: tell the user; installing is their call. Other codes → [error codes](references/protocol-error-codes.md).
Compile status, events, analytics → [observability](references/protocol-observability.md); closed Editor + opt-in Unity CLI → [unity-cli](references/protocol-unity-cli.md).
Windows (PowerShell / Git Bash): [windows](references/windows.md) rewrites these calls; or use `scripts/unity_skills.py`.
讨论
提问、分享技巧或报告无法使用。按“有帮助”数量排序。
来自 GitHub
- GitHub 星标
- 1,949
- 开发者
- Besty0728
- 许可证
- MIT
- 最近代码更新
- 2026年11月15日
- repo 中的 SKILL.md 文件(含译本和模板)
- 2
- 主题
- ai, antygravity-ai, chatgpt, claude, claudecode, codex, cursor, gemini-cli, geminicli, harness, mr, opencode