Unity-MCP connects AI coding agents to the Unity game engine through MCP, letting them work with the Unity Editor and running games. It is used by game developers to generate code, automate development and testing, debug projects, and add AI-driven behavior to games. The catalogue entries provide skills, instructions, an MCP integration, and settings for operating this workflow.
Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add IvanMurzak/Unity-MCP --skill rql-mutategit clone --depth 1 https://github.com/IvanMurzak/Unity-MCPWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/ivanmurzak/unity-mcp/rql-mutate)<a href="https://agentmods.dev/skills/ivanmurzak/unity-mcp/rql-mutate"><img src="https://agentmods.dev/badge/skills/ivanmurzak/unity-mcp/rql-mutate/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/ivanmurzak/unity-mcp/rql-mutate"><img src="https://agentmods.dev/badge/skills/ivanmurzak/unity-mcp/rql-mutate.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00038 | $0.01259 |
| Opus 5.5 | $0.00015 | $0.00504 |
| Sonnet 5.5 | $0.00008 | $0.00252 |
| Haiku 4.5 | $0.00004 | $0.00126 |
Grade A, and why
rql-mutate scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured yesterday.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 96 lines — stays where its author put it; the contents beside it link to each section on GitHub.
RQL / Mutate
Change the live Unity Editor scene with an RQL query whose last stage is set or call. All-or-nothing: every target and value is checked before anything is written; a failure changes nothing. Undoable in the Editor (Ctrl+Z). Changed assets, scenes and the open prefab are saved to disk automatically; an untitled scene needs SaveAs(path). dry_run=true returns the receipt ($ref, member, old, new) without changing anything. Do exactly the change requested; a read-only member fails with readonly-member instead of being approximated. set path = expr, ... assign members (paths may go through structs: set transform.position.y = transform.position.y + 1) call Method(args) invoke a method the tools expose Examples: objects | where name == "Player" | set transform.position = "(0, 1, 0)" objects | get components | where $ is Light | set intensity = intensity * 2, color = "#ffcc88" objects | where tag == "Enemy" | set active = false unity.asset("Assets/Mats/Red.mat") | set properties["_Color"] = "(1, 0, 0, 1)" objects | where name == "Spawner" | set targets = append(targets, ref("GameObject#1234")) Create / add / remove (call; the receipt's result column holds the new object's $ref for follow-up queries): editor | call CreateGameObject("Props") new root GameObject editor | call CreateGameObject("Crate", ref("GameObject#1234"), "Cube") child of a parent, optional primitive objects | where name == "Crate" | call AddComponent("Rigidbody") add a component by type name objects | where name == "Old" | call Destroy() remove a GameObject (components: get components | where $ is X | call Destroy()) objects | where name == "Lamp" | set transform.parent = ref("GameObject#1234") reparent (keeps the world position) objects | where name == "Crate" | call Duplicate() copy next to it (returns the copy) Scenes, prefabs, assets, Editor state: editor | call OpenScene("Assets/Scenes/Level.unity") editor | call NewScene("Assets/Scenes/New.unity") scenes | where path == "" | call SaveAs("Assets/Scenes/New.unity") scenes | where name == "Level" | call SetActive() editor | call InstantiatePrefab("Assets/Prefabs/Enemy.prefab", ref("GameObject#1234"), "(0, 0, 5)") objects | where name == "Crate" | call SaveAsPrefab("Assets/Prefabs/Crate.prefab") editor | call OpenPrefab("Assets/Prefabs/Enemy.prefab") then prefab | recurse children | … then editor | call ClosePrefab() editor | call CreateMaterial("Assets/Mats/Glow.mat") editor | call MoveAsset("Assets/A.mat", "Assets/Mats/A.mat") editor | set selection = [ref("GameObject#1234")] editor | set isPlaying = true profiler | call Start() profiler | call Save("Logs/profile.json") (read: profiler | select memory frame) Object literals build values: set settings = {speed: 2, "$type": "FastSettings"}. At most 20 targets per call.
Query part: the RQL of rql-query (full guide and the sources in its description); unity.assets("t:Material"), unity.asset(path), unity.find("Name"), ref("Light#1234"). Stages: where, select, sort, first N, get , recurse children. Operators: == != < > and or not, contains, starts-with, $ is Light. Paths a.b.c, list[0], map["key"]. Vectors/colors as "(1, 2, 3)" or "#ff8800"; enums as strings.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- yesterday First seen · 96 lines · 38 tokens per session scan A bf3d1c411a24
rql-mutate is a skill published in the GitHub repository IvanMurzak/Unity-MCP (4,483 stars, last pushed today), licensed Apache-2.0. It adds 38 tokens to every session and 1,259 once invoked, about $0.0002 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-10-16.
Other skills, from other repositories
unity-addressables
Use Addressables with explicit grouping, loading, update, and fallback rules instead of ad hoc content loading.
unity-build-release
Run Unity builds, CI, and release packaging with reproducible configuration and platform-aware checks.
unity-csharp-standards
Apply Unity-specific C# standards that preserve readability, testability, and safe engine integration.
unity-editor-tooling
Create Unity editor tools that make content production safer, faster, and less dependent on tribal knowledge.
unity-gameplay-patterns
Use common Unity gameplay architecture patterns without hiding ownership inside scenes or inspector wiring.
unity-input-system
Implement Unity input with explicit action maps, device support, and UI/gameplay separation.