Unity Skills logo

Unity Skills

CommunityPopular
Besty0728
unity-skills

Automate the Unity Editor through a local REST API — create and edit scripts, build scenes and prefabs, manage assets/materials/lighting, run tests, and drive hundreds of Editor operations across modules. Use when the user wants to actually operate the Unity Editor from chat — create or modify GameObjects/scripts/scenes/assets, batch-edit, or run Editor automation — in any language. Not needed for conceptual Unity Q&A that touches no Editor state — read the matching advisory doc under skills/ instead. 当用户要从对话里实际操作 Unity 编辑器(创建/修改/批量编辑/运行测试)时使用,任何语言均可触发;纯概念问答无需本协议。

Overview

PublisherBesty0728
RepositoryUnity-Skills
Skill nameunity-skills
Stars
1.8K
Forks
164
Bundled files
92
LicenseMIT
Links
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • 92 bundled files

    Scripts, templates, and references the model can read while it works. Files are read-only and never executed.

  • Open source

    Published by Besty0728 on GitHub. Read the source before you install it.

Installation

Install the Unity Skills AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.

Use it in TypingMind

Enable Unity Skills in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Unity Skills on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Unity Skills is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

Unity Skills

Module and reference docs are English-only. Match requests to modules by meaning in any language, and always reply in the user's language.

Route first: automate, or guide the user?

  • Automate — you call REST skills and drive the Editor yourself. Stay here when the user asked you to do it, or the task needs traversal, search, batch, exact numeric values, or consistency across many objects.
  • Guide — you name the menus and Inspector fields and write nothing through REST. Switch to SKILL_GUIDE.md (guidance boundary table + manual-* routing) when the user asked how to do it themselves, or /health reports a surfaceProfile other than full.

First Contact Checklist

Before the first skill call in a session:

  1. GET /health — discover the server (ports 80908100); read currentMode ("approval" / "auto" / "bypass"), panelApprovalRequired, pendingCount, and surfaceProfile (full / guide / noSceneAuthoring). surfaceProfileHint is text only when the profile is not full, else null.
  2. Branch on currentMode: under approval the first write to any FullAuto skill returns MODE_RESTRICTED and needs a grant; under auto/bypass writes execute directly (self-assess risk under auto). Grants are single-shot; the permanent form is the user-managed Allowlist. Gates, grant protocol and mode table → operating mode.
  3. GET /skills/meta — once per session: the constants shared by every skill (categories, operationTypes, reservedBodyParameters, workflowTrackedSkills, schemaVersion, defaults). Read them here, not out of each manifest, then start discovery.

Schema: pick the cheapest layer

Server-cached (ETag/304), off the main thread; send Accept-Encoding: gzip.

LayerEndpointSizeUse when
Default: start hereGET /skills/recommend?intent=...&includeSchema=true4–14 KBOne intent; scored candidates + schemas. topN caps how many (default 10), wire=v2 about halves it.
directory + categoryGET /skills then GET /skills/schema?category=<Category>~19 KB + 13–44 KBTouches one or two areas.
summaryGET /skills?summary=1~143 KBExploratory / cross-module, or cheaper layers left you unsure.
full schemaGET /skills/schema~618 KBRare; many modules' signatures at once.

Bare GET /skills is the brief directory — names by category only; the old full listing needs ?full=1.

Module docs and the scoped schema are complements: schema has exact signatures; the module doc has guardrails, return shapes and traps schema omits. On an unfamiliar module read both.

Wire format v2 (?wire=v2)

?wire=v2 slims per-skill entries on ?full=1, /skills/schema (full or scoped), a filtered /skills, and recommend — never bare GET /skills (always the brief directory). v1 is default:

  • A flags array replaces v1's six booleans and adds v2-only longRunning: readOnly, tracksWorkflow, mutatesScene, mutatesAssets, mayTriggerReload, mayEnterPlayMode, longRunning. An absent flag is false.
  • Omitted means default: riskLevel appears only when not "low", supportsDryRun only when false, null members are dropped — an absent key never reads as null.
  • Every v2 response carries a defaults block — take the rule from the payload, don't memorize it.

Execute: batch, dryRun gate, anti-hallucination rules

POST /skills/batch — up to 50 steps per call ({"steps":[{"skill","args"}],"continueOnError":false}; ?mode=dryRun validates every step): the largest round-trip saving available → batch.

Before executing any skill whose exact parameters you don't already hold, dryRun it: POST /skill/<name>?mode=dryRun. Iterate until valid: true, then execute without ?mode=dryRun. valid: true means the four validation error buckets are empty — warnings never block, and target existence is never checked. The top-level authorization ({allowed, blockedBy, currentMode, allowlisted, hint}, plus surfaceProfile on a SURFACE_EXCLUDED block) previews interception: blockedBy is MODE_RESTRICTED, MODE_FORBIDDEN, SURFACE_EXCLUDED, or null when the call would run — settle grants and exclusions there, not on the real call.

After a write, verify in the Editor, not from the echo: *_get_info / *_get_properties, or find_objects_by_name.

RuleRequirement
Uncertain parametersMust dryRun first. Never guess parameters from a skill name.
Skill name mismatchIf the name does not appear in schema/recommend results, do not invent it.
Call failureRead suggestedFixes; when pointed to a module doc, actually read it.

Surface profile

surfaceProfile says which slice of the skill surface the user exposed. You cannot change it — the user switches it in the UnitySkills panel. (guideMode is legacy; surfaceProfile is authoritative.)

ProfileHiddenWhat to do
fullnothingNormal automation.
guidewrite skills in GameObject, Component, Material, Scene and SampleGive manual steps via SKILL_GUIDE.md and the manual-* docs. Read-only skills there still work, as does every other module.
noSceneAuthoringevery scene-authoring write, incl. any mutatesScene skillDo the rest of the task normally; if it genuinely needs scene authoring, say so and let the user switch back to full.

Calling a hidden skill returns SURFACE_EXCLUDED, and the response names the document to read (or the profile to leave). It is a configuration boundary, not a failure: never retry it, never route around it through another module.

Module routing

Every skill is named module_verb, so the prefix is the routing key. GET /skills/recommend?intent=... ranks candidates, or read the index — all 80 modules with mode labels, incl. docs-only ones (manual-*, *-design, unity-cli) that define no REST skills → module index.

Error codes quick reference

CodeMeaningPointer
MODE_RESTRICTED / MODE_FORBIDDENNeeds a user grant / needs Bypass or Allowlist.operating mode
SURFACE_EXCLUDEDHidden by the current surfaceProfile."Surface profile" above; read the doc named in the response.
MISSING_PARAM / TARGET_NOT_FOUNDBad or unresolvable arguments.dryRun for the schema; locate the target first.
any other codeerror codes — transient ones (COMPILING/RATE_LIMIT/QUEUE_FULL/SERVER_STOPPED) are auto-retried by the Python client.

Observability and Unity CLI pointers

Compilation status, events, analytics → observability. Unity CLI cold start (opt-in, v2.3+) → unity-cli.

Current snapshot: 805 REST skills, 56 source files, 54 categories, 82 module documentation directories (54 REST/module docs + 28 advisory docs), Unity 2022.3+, default timeout 15 minutes.

Python helper: unity-skills/scripts/unity_skills.py

Bundled files

The model reads these on demand while the skill is loaded. They are exposed as readable files and are never executed.

and 32 more files.

Frequently asked questions

What does the Unity Skills AI skill do?

Automate the Unity Editor through a local REST API — create and edit scripts, build scenes and prefabs, manage assets/materials/lighting, run tests, and drive hundreds of Editor operations across modules. Use when the user wants to actually operate the Unity Editor from chat — create or modify GameObjects/scripts/scenes/assets, batch-edit, or run Editor automation — in any language. Not needed for conceptual Unity Q&A that touches no Editor state — read the matching advisory doc under skills/ instead. 当用户要从对话里实际操作 Unity 编辑器(创建/修改/批量编辑/运行测试)时使用,任何语言均可触发;纯概念问答无需本协议。

Why use Unity Skills on TypingMind?

Because you install it once and use it with any model. Unity Skills is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Unity Skills in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/Besty0728/Unity-Skills/tree/main/SkillsForUnity/unity-skills~. TypingMind reads its SKILL.md and bundles its files and installs it as a skill you can enable per chat.

Which AI models can use Unity Skills?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Unity Skills?

As many as you like. As long as a model supports skills, you can use Unity Skills with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Unity Skills AI skill free?

Yes. It is published on GitHub by Besty0728 under the MIT license. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇