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 agentmods add instructions/coplaydev/unity-mcp/claude-mdgit clone --depth 1 https://github.com/CoplayDev/unity-mcpWhat 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 | $0.02261 | $0.02261 |
| Opus 5 | $0.01130 | $0.01130 |
| Sonnet 5 | $0.00452 | $0.00452 |
| Haiku 4.5 | $0.00226 | $0.00226 |
Grade A, and why
unity-mcp CLAUDE.md 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 — 201 lines — stays where its author put it; the contents beside it link to each section on GitHub.
CLAUDE.md
This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.
What This Project Is
MCP for Unity is a bridge that lets AI assistants (Claude, Cursor, Windsurf, etc.) control the Unity Editor through the Model Context Protocol (MCP). It enables AI-driven game development workflows - creating GameObjects, editing scripts, managing assets, running tests, and more.
Architecture
AI Assistant (Claude/Cursor)
↓ MCP Protocol (stdio/HTTP)
Python Server (Server/src/)
↓ WebSocket + HTTP
Unity Editor Plugin (MCPForUnity/)
↓ Unity Editor API
Scene, Assets, Scripts
Two codebases, one system:
Server/- Python MCP server using FastMCPMCPForUnity/- Unity C# Editor package
Three Layers on the Python Side
The Python server has three distinct layers. These are not auto-generated from each other:
| Layer | Location | Framework | Purpose |
|---|---|---|---|
| MCP Tools | Server/src/services/tools/ |
FastMCP (@mcp_for_unity_tool) |
Exposed to AI assistants via MCP protocol |
| CLI Commands | Server/src/cli/commands/ |
Click (@click.command) |
Terminal interface for developers |
| Resources | Server/src/services/resources/ |
FastMCP (@mcp_for_unity_resource) |
Read-only state exposed to AI assistants |
MCP tools call Unity via WebSocket (send_with_unity_instance). CLI commands call Unity via HTTP (run_command). Both route to the same C# HandleCommand methods.
Transport Modes
- Stdio: Single-agent only. Separate Python process per client. Legacy TCP bridge to Unity. New connections stomp old ones.
- HTTP: Multi-agent ready. Single shared Python server. WebSocket hub at
/hub/plugin. Session isolation viaclient_id.
Code Philosophy
1. Domain Symmetry
Python MCP tools mirror C# Editor tools. Each domain exists in both:
Server/src/services/tools/manage_material.py↔MCPForUnity/Editor/Tools/ManageMaterial.cs- CLI commands (
Server/src/cli/commands/) also mirror these but are a separate implementation.
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 · 201 lines · 2,261 tokens per session scan A ddf38c904a14
unity-mcp CLAUDE.md is an instructions file published in the GitHub repository CoplayDev/unity-mcp (13,781 stars, last pushed 24d ago), licensed MIT. It adds 2,261 tokens to every session, about $0.0113 per session on Opus 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 2026-08-30.
Other instructions, from other repositories
mcp-unity AGENTS.md
Instructions for CoderGamester/mcp-unity, covering mcp unity — ai agent guide (mcp package), purpose (what this repo is), how it works (high-level data flow), key defaults & invariants and repo layout (where to change what).
mcp-unity CLAUDE.md
Instructions for CoderGamester/mcp-unity, a project described as: Model Context Protocol (MCP) plugin to connect with Unity Editor — designed for Cursor, Claude Code, Codex, Windsurf and other IDEs.
Unity-MCP copilot-instructions.md
Instructions for IvanMurzak/Unity-MCP, covering project guidelines, critical rules, no c# reflection for private access, code review and code style.
gortex CLAUDE.md
Instructions for zzet/gortex, covering gortex, build & test, codebase overview, discovery (read once, then keep using) and llm provider (powers ask and searchsymbols assist: modes).
rails-ai-context AGENTS.md
Instructions for crisnahine/rails-ai-context, covering rails-ai-context, agent skills, issue tracker, triage labels and domain docs.
inspecto CLAUDE.md
Instructions for inspecto-dev/inspecto, covering inspecto — claude code development guide, project overview, monorepo structure, development phases (load each file as needed) and key architectural decisions.