unityctl

English | 한국어

The execution layer for AI-driven game development.

Give your AI agent 133 commands to build Unity scenes, write C# scripts, validate builds, and ship games — with automatic rollback when things go wrong.

133 CLI commands · 12 MCP tools · 689 tests · Windows / macOS / Linux

---

The Problem

AI agents can write code, but they can't build games — because Unity has no programmatic interface for scene editing, asset management, or project validation.

Existing Unity MCP servers try to fix this, but they create new problems for AI agents:

Pain Point	Impact on AI Agent
45 KB+ schemas loaded every turn	Wastes tokens on tool definitions instead of reasoning
No validation feedback	Agent can't tell if the scene is broken after changes
No rollback	One bad command corrupts the project state
WebSocket drops on Play Mode	Agent loses connection during Unity's Domain Reload
Editor must be open	CI/CD pipelines can't run without a GUI

The Solution

unityctl is a .NET CLI + MCP server that turns Unity Editor into a programmable API.

For AI agents, this means a closed-loop automation cycle — the agent doesn't just execute commands, it can verify results, diagnose failures, and recover from mistakes:

Other tools give agents hands. unityctl gives agents hands, eyes, and a safety net.

---

What AI Agents Can Build

Scene Construction

"Create a platformer level with a floor, walls, and a player spawn point"

# Agent creates scene structure
unityctl scene create --path "Assets/Scenes/Level01.unity" --project $P
unityctl mesh create-primitive --type Plane --name "Floor" --scale "[10,1,10]" --project $P
unityctl mesh create-primitive --type Cube --name "Wall" --position "[5,1,0]" --scale "[0.5,2,10]" --project $P
unityctl gameobject create --name "PlayerSpawn" --project $P
unityctl component add --id "<PlayerSpawnId>" --type "BoxCollider" --project $P

# Agent verifies the scene
unityctl scene hierarchy --project $P --json      # check structure
unityctl screenshot capture --project $P           # visual verification
unityctl project validate --project $P --json      # camera? lights? errors?

Script Authoring with Compile Verification

"Write a player movement script and make sure it compiles"

# Agent writes code
unityctl script create --path "Assets/Scripts/PlayerMovement.cs" --className "PlayerMovement" --project $P
unityctl script patch --path "Assets/Scripts/PlayerMovement.cs" \
  --startLine 8 --insertContent "public float speed = 5f;" --project $P

# Agent checks compilation — and fixes errors in a loop
unityctl script validate --project $P --wait       # trigger recompile
unityctl script get-errors --project $P --json     # structured CS errors
# if errors: read error, patch fix, validate again

Safe Batch Operations with Rollback

"Set up physics layers for Player, Enemy, and Projectile — roll back if anything fails"

unityctl batch execute --project $P --rollbackOnFailure true --commands '[
  {"command": "layer-set", "parameters": {"index": 8, "name": "Player"}},
  {"command": "layer-set", "parameters": {"index": 9, "name": "Enemy"}},
  {"command": "layer-set", "parameters": {"index": 10, "name": "Projectile"}},
  {"command": "physics-set-collision-matrix", "parameters": {"layer1": 10, "layer2": 10, "ignore": true}}
]'
# If any command fails, all changes are automatically rolled back via Undo

Build Verification Pipeline

"Check if the project is ready to ship"

# Agent reads the failure, fixes it, validates again
unityctl gameobject create --name "Main Camera" --project $P
unityctl component add --id "<MainCameraId>" --type "Camera" --project $P
unityctl gameobject set-tag --id "<MainCameraId>" --tag "MainCamera" --project $P
unityctl project validate --project $P --json   # valid: true

---

What To Build First

If you want to prove unityctl in public, do not start with Minecraft. Start with a showcase ladder that matches the strongest verified loops in the current toolchain:

1. Zero-to-playable: a small 3D arena microgame built from primitives, scripts, UI, physics, and verification artifacts.
2. Vertical slice: expand that into a polished top-down survival or base-defense prototype with prefabs, NavMesh, materials, audio, and build validation.
3. Sandbox step: only then move into chunked worlds, crafting, procedural terrain, and save-heavy systems.

The best first showcase for unityctl is a small 3D survival / base-defense game, not an open-world sandbox. It is easy to understand in screenshots and GIFs, maps cleanly to scene editing + script patching + rollback + visual verification, and can grow into a more complex sandbox later.

See Showcase Roadmap for:

• the recommended public demo scope
• the asset checklist to prepare before building
• the pre-production plan for using unityctl effectively
• the gaps worth closing before attempting a Minecraft-like demo

---

Why unityctl for AI Agents?

	unityctl	Existing Unity MCP
Schema overhead	5 KB per session (9x smaller)	45 KB+ loaded every turn
Validation loop	project validate + scene diff + screenshot capture	Agent flies blind
Error recovery	script get-errors with file/line/column	Raw console output or nothing
Safe experimentation	batch execute --rollbackOnFailure + undo	No rollback — mistakes are permanent
Connection stability	Named Pipe — survives Domain Reload	WebSocket drops, reconnect needed
CI/CD	check / test / build --dry-run work headless	Editor must be open
Diagnostics	doctor classifies failures + suggests next steps	"Connection failed"
Commands	133 (read + write + validate + diagnose)	~34-200 tools
Audit trail	NDJSON flight recorder for every command	No history
Runtime	Native .NET — no Python/TS bridge	Bridge overhead
Install	dotnet tool install -g unityctl	Node.js + npm + port config
License	MIT	Varies

Token Efficiency

AI agent costs are dominated by tool schemas sent every turn. unityctl uses on-demand schema loading:

The 12 MCP tools cover the full 131-command surface through unityctl_query (read), unityctl_run (write), and unityctl_schema (lookup).

Measured: Claude Code Token Cost (2026-03-20)

When Claude Code runs 5 read-only QA operations (compile check, scene hierarchy, robot catalog, DH table, build settings), the cumulative token cost differs dramatically:

Stack	Schema (once)	5 ops x 1	5 ops x 10
unityctl via Bash	0 tok	1,780 tok	17,800 tok
unityctl MCP (12 tools)	1,256 tok	2,957 tok	18,261 tok
CoplayDev MCP (30 tools)	11,427 tok	12,158 tok	18,742 tok

Key findings:

• unityctl via Bash has zero schema overhead — the Bash tool is already in Claude Code's system prompt, so no additional tokens are spent on tool definitions
• CoplayDev MCP loads 45 KB of schemas (30 tools), but only 1 out of 5 QA operations has a matching tool
• In a typical short session, unityctl via Bash uses 6.8x fewer tokens than CoplayDev MCP
• Full benchmark methodology and raw data: docs/benchmark/

---

Selection-aware Routing

# Pin the current Unity project once
unityctl editor select --project /path/to/project

# Or pin a running Unity PID when it maps to a single project
unityctl editor select --pid 55028

# Inspect the current selection
unityctl editor current --json

# See running Unity instances with PID / project / IPC status
unityctl editor instances --json

# These CLI commands can now omit --project
unityctl ping --json
unityctl status --json
unityctl check --json
unityctl doctor --json

# Run a small verification bundle (artifacts-first)
unityctl workflow verify --file verify.json --project /path/to/project --json

Install

# CLI (requires .NET 10 SDK)
dotnet tool install -g unityctl

# MCP server for AI agents
dotnet tool install -g unityctl-mcp

Bootstrap notes:

• --source accepts a local Unityctl.Plugin folder or a Git URL: https://github.com/kimjuyoung1127/unityctl.git?path=/src/Unityctl.Plugin#v0.3.2
• GitHub Release CLI archives are framework-dependent (not self-contained) today.

Apple Silicon macOS Validation

Manual validation was completed on an Apple silicon MacBook Air using Homebrew, .NET SDK 10.0.105, Unity Hub, and Unity editors 6000.0.64f1 and 6000.3.11f1.

Validated path:

• dotnet tool install -g unityctl
• dotnet tool install -g unityctl-mcp
• unityctl editor list
• unityctl init --project <project> --source /path/to/unityctl/src/Unityctl.Plugin
• unityctl ping --project <project> --json
• unityctl doctor --project <project> --json
• unityctl status --project <project> --json
• unityctl check --project <project> --json

Observed result on a Unity 6000.0.64f1 project: ping returned pong, doctor reported IPC connected, status returned Ready, and check passed on macOS.

Project compatibility note: if a Unity project or third-party package is pinned to Unity 6.0 LTS, opening that same project in 6000.3+ can fail before unityctl is involved. During validation, reopening the project in its pinned 6000.0.64f1 editor resolved the project-side render pipeline error.

Quick Start

# 1. Install the Editor plugin
unityctl init --project /path/to/project \
  --source "https://github.com/kimjuyoung1127/unityctl.git?path=/src/Unityctl.Plugin#v0.3.2"

# 2. Open the project in Unity Editor, then verify connectivity
unityctl ping --project /path/to/project --json
unityctl status --project /path/to/project --json

# 3. Start building
unityctl gameobject create --name "Player" --project /path/to/project
unityctl component add --id "<PlayerId>" --type "Rigidbody" --project /path/to/project
unityctl scene save --project /path/to/project

# 4. Validate
unityctl project validate --project /path/to/project --json

# 5. Build
unityctl build --project /path/to/project --dry-run    # 13 preflight checks

MCP Setup (AI Agents)

Add to your Claude Code / Cursor / VS Code MCP config:

{
  "mcpServers": {
    "unityctl": {
      "command": "unityctl-mcp"
    }
  }
}

12 MCP Tools

Tool	Type	Description
unityctl_query	Read	Unified read: asset, gameobject, scene, component, UI, physics, lighting, tags
unityctl_run	Write	Unified write: create, delete, modify, script, material, prefab, batch
unityctl_schema	Meta	On-demand parameter lookup (by command or category)
unityctl_build	Action	Build player with 13 preflight checks
unityctl_check	Action	Compile verification (headless)
unityctl_test	Action	EditMode / PlayMode tests
unityctl_exec	Action	Execute arbitrary C# expression
unityctl_status	Read	Editor state + connectivity
unityctl_ping	Read	Fast connectivity check
unityctl_watch	Stream	Real-time console / hierarchy / compilation events
unityctl_log	Read	Flight recorder query
unityctl_session_list	Read	Active session list

---

Commands (133)

Core (13)

Command	Description
ping	Check Unity connectivity
status	Editor state (with --wait smart polling for Domain Reload)
check	Verify script compilation (headless)
build	Build player with --dry-run preflight (13 checks)
test	Run EditMode / PlayMode tests
doctor	Diagnose connectivity + suggest recovery steps
project validate	Game readiness check (compile, scenes, camera, lights, console, editor)
init	Install plugin to Unity project
editor list	Discover installed Unity editors
editor instances	List running Unity Editor instances
editor current	Show the selected Unity project target
editor select	Select a Unity project target, or a unique running PID, for project-less CLI routing
workflow verify	Run artifact-first verification steps (projectValidate, capture, imageDiff, consoleWatch, uiAssert, playSmoke)

Scene & GameObject (19)

Command	Description
scene snapshot	Capture scene state
scene hierarchy	Scene hierarchy tree
scene diff	Property-level scene diff with epsilon
scene save	Save active scene
scene open	Open scene by path
scene create	Create new scene
gameobject create	Create GameObject
gameobject delete	Delete GameObject
gameobject rename	Rename GameObject
gameobject move	Reparent GameObject
gameobject find	Find by name/tag/component
gameobject get	Get GameObject details
gameobject set-active	Toggle active state
gameobject set-tag	Set tag
gameobject set-layer	Set layer
component add	Add component
component remove	Remove component
component get	Get component properties
component set-property	Set component property

Assets & Materials (21)

Command	Description
asset find	Search by type/label/path
asset get-info	Asset metadata
asset get-dependencies	Direct dependencies
asset reference-graph	Reverse-reference graph
asset create	Create asset
asset create-folder	Create folder
asset copy	Copy asset
asset move	Move/rename asset
asset delete	Delete asset
asset import	Reimport asset
asset refresh	Refresh AssetDatabase
asset get-labels	Get labels
asset set-labels	Set labels
material create	Create material
material get	Get material properties
material set	Set material property
material set-shader	Change shader
prefab create	Create prefab from GameObject
prefab unpack	Unpack prefab instance
prefab apply	Apply prefab overrides
prefab edit	Enter/exit prefab edit mode

Scripting & Code Analysis (10)

Command	Description
script create	Create C# script from template
script edit	Replace script content (whole-file)
script patch	Line-level insert/delete/replace
script delete	Delete script file
script validate	Trigger compilation and verify
script list	List MonoScript assets
script get-errors	Structured compile errors (file/line/column/code)
script find-refs	Find symbol references across all scripts
script rename-symbol	Rename symbol across all scripts (with --dry-run)
exec	Execute C# expression in Unity

Editor Control (18)

Command	Description
play start/stop/pause	Start, stop, or pause play mode
editor pause	Toggle editor pause
editor focus-gameview	Focus Game View
editor focus-sceneview	Focus Scene View
player-settings get/set	PlayerSettings read/write
project-settings get/set	Project settings read/write
console clear	Clear console
console get-count	Log/warning/error counts
define-symbols get/set	Scripting define symbols
tag list/add	Tag management
layer list/set	Layer management
undo	Undo last operation
redo	Redo last undone operation

Build & Deployment (6)

Command	Description
build-profile list/get-active/set-active	Build profile management
build-target switch	Switch build platform
build-settings get-scenes/set-scenes	Build scene list

Physics, Lighting & NavMesh (12)

Command	Description
physics get-settings/set-settings	DynamicsManager
physics get-collision-matrix/set-collision-matrix	32x32 layer collision
lighting bake/cancel/clear	Lightmap baking
lighting get-settings/set-settings	Lightmap settings
navmesh bake/clear/get-settings	NavMesh

UI & Mesh (8)

Command	Description
ui canvas-create	Create UI Canvas
ui element-create	Create Button, Text, Image, etc.
ui set-rect	Set RectTransform
ui find	Find UI elements
ui get	Get UI element details
ui toggle	Set Toggle state
ui input	Set InputField text
mesh create-primitive	Create Cube/Sphere/Plane/Cylinder/Capsule/Quad

Automation & Monitoring (15)

Command	Description
batch execute	Transaction with rollback
workflow run	JSON workflow execution
watch	Real-time event streaming
log	Flight recorder query
session list/stop/clean	Session management
screenshot	Scene/Game View capture (base64)
schema / tools	Machine-readable metadata
package list/add/remove	Package management
animation create-clip/create-controller	Animation assets

---

Architecture

AI Agent (LLM)                unityctl-mcp              unityctl CLI             Unity Editor
Claude / GPT / Gemini         12 MCP tools              133 commands             Plugin (IPC)
        |                          |                          |                       |
        |--- MCP (stdio) -------->|                          |                       |
        |                          |--- CLI invocation ----->|                       |
        |                          |                          |--- IPC (~100ms) ---->|
        |                          |                          |    or Batch (30s+)   |
        |                          |                          |<--- JSON response ---|
        |                          |<--- result -------------|                       |
        |<--- tool result --------|                          |                       |

unityctl.slnx
+-- src/Unityctl.Shared   (netstandard2.1)  Protocol + models
+-- src/Unityctl.Core     (net10.0)         Business logic
+-- src/Unityctl.Cli      (net10.0)         CLI shell
+-- src/Unityctl.Mcp      (net10.0)         MCP server
+-- src/Unityctl.Plugin   (Unity UPM)       Editor bridge (IPC server)
+-- tests/*                                 633 xUnit tests

---

Platforms

Platform	CLI	IPC Transport	Batch	CI
Windows	✅	Named Pipe	✅	✅
macOS	✅	Unix Domain Socket	✅	✅
Linux	✅	Unix Domain Socket	✅	✅

Prerequisites

• .NET 10 SDK
• Unity 2021.3+

Terminal Output

Documentation

• Getting Started — installation, setup, and common workflows
• AI Agent Quickstart — MCP setup and agent integration guide
• Showcase Roadmap — recommended demo game ladder, asset checklist, and pre-production plan
• Architecture — system design and transport diagrams
• Glossary — key terms and concepts

Changelog

See GitHub Releases for version history.

License

MIT — see LICENSE