Borrowing it
Nothing to install: this file belongs to IvanMurzak/Unreal-MCP. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.
curl -O https://raw.githubusercontent.com/IvanMurzak/Unreal-MCP/main/AGENTS.mdgit clone --depth 1 https://github.com/IvanMurzak/Unreal-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/instructions/ivanmurzak/unreal-mcp/agents-md)<a href="https://agentmods.dev/instructions/ivanmurzak/unreal-mcp/agents-md"><img src="https://agentmods.dev/badge/instructions/ivanmurzak/unreal-mcp/agents-md/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/instructions/ivanmurzak/unreal-mcp/agents-md"><img src="https://agentmods.dev/badge/instructions/ivanmurzak/unreal-mcp/agents-md.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 | $0.02888 | $0.02888 |
| Opus 5.5 | $0.01155 | $0.01155 |
| Sonnet 5.5 | $0.00578 | $0.00578 |
| Haiku 4.5 | $0.00289 | $0.00289 |
Grade A, and why
Unreal-MCP AGENTS.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 4d ago.
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 — 124 lines — stays where its author put it; the contents beside it link to each section on GitHub.
AGENTS.md — Unreal-MCP
Unreal-MCP connects MCP clients (Claude, Cursor, Copilot, …) to Unreal Engine
— the Unreal sibling of Unity-MCP and Godot-MCP. Editor mode (default) drives the Unreal Editor;
opt-in runtime mode drives a running PIE / Standalone / packaged Development build. Status: beta.
The MCP server is the shared GameDev-MCP-Server
(gamedev-mcp-server); no server code lives here — server changes go to that repo.
docs/ARCHITECTURE.md is authoritative (defer to it on conflict; read the
relevant section before non-trivial work): IPC §1, tool registration §2, schema §3, dispatcher §4,
extensions §5, sidecar §6, UI §7, config §8, repo/CI §9, roadmap §10, risks §11, runtime §12, prompts &
resources §A. Other docs: README.md (users; canonical tool inventory),
docs/RELEASING.md, docs/EXTENSIONS.md (extension
authors), docs/RUNTIME-E2E.md (in-game walkthrough).
Architecture
Unreal is C++ and cannot host the .NET McpPlugin/ReflectorNet/SignalR stack in-process, so it runs in an
auto-managed .NET 9 sidecar (unreal-mcp-bridge, bridge/), a local child of whatever loads the
plugin (Editor, or the game in an opt-in packaged build). The C++ plugin listens on loopback TCP; the
sidecar dials it with a one-shot token sent over stdin (never argv, never logged — §1.4) and
relays IPC ⇄ SignalR to the MCP server (cloud ai-game.dev by default, or a local gamedev-mcp-server).
Do not conflate the two child processes: unreal-mcp-bridge (always, when connected) and
gamedev-mcp-server (Custom/local mode only; downloaded by the CLI, pinned by
cli/src/lib/server-version.ts SERVER_VERSION, overridden by UNREAL_MCP_SERVER_PATH; the editor can
Start/Stop it via FUnrealMcpServerManager, Custom + http only).
Runtime mode (§12). Locked decisions: reuse the sidecar (no in-process C++ server); runtime infra +
ping live in UnrealMcpRuntime; a shipped game never auto-connects. UUnrealMcpRuntimeSubsystem
(a UGameInstanceSubsystem) only arms the listener; a connection is an explicit
Connect(Host, Token, Mode, bAllowRemoteHost) (C++, Blueprint, or UnrealMcp.Connect console command).
The runtime built-in surface is ping only (a system tool, so tools/list is legitimately empty;
the UnrealMcp.RuntimeSubsystem spec asserts the registry is exactly {ping}); all engine-development
tools are editor-only, and games bring their own via IUnrealMcpToolProvider. Security gates, all in
Connect() (§12.8): opt-in only; kill switch UUnrealMcpRuntimeSettings::bRuntimeMcpEnabled
(Project Settings → Plugins → Unreal MCP (Runtime), DefaultGame.ini, default false); Shipping gate bUnrealMcpAllowShipping in *.Build.cs
(default false → UNREAL_MCP_ALLOW_SHIPPING=0, also controls staging the sidecar into Shipping);
loopback host unless bAllowRemoteHost=true; loopback IPC + stdin token. Desktop only (Win64/Mac/Linux).
Consumers exclude the plugin from packaged games with a TargetDenyList: ["Game","Client","Server"] in
their own .uproject plugin reference (not our .uplugin); any game-module dependency on an
UnrealMcp* module overrides that.
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.
- 4d ago Changed · -1 tokens per session 38fe8e55539a
- 8d ago Changed · +120 tokens per session 67180a9753b0
- 9d ago Changed · +2 lines · +59 tokens per session 39fe3ae4f4dd
- 18d ago Changed · +12 tokens per session f0e8bb31ddae
- 19d ago Changed · +2 lines · +86 tokens per session 66c3af71b7ad
- 20d ago Changed · +3 lines · +108 tokens per session e71ef9a7ed1b
- 21d ago Changed · +4 lines · +86 tokens per session 888c915b37ad
- 25d ago Changed · +1 lines · +28 tokens per session 626927de2d9d
- 26d ago First seen · 112 lines · 2,390 tokens per session scan A ad485f97b5c5
Unreal-MCP AGENTS.md is an instructions file published in the GitHub repository IvanMurzak/Unreal-MCP (57 stars, last pushed yesterday), licensed Apache-2.0. It adds 2,888 tokens to every session, about $0.0116 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 2026-10-12.
Unreal-MCP AGENTS.md compared with instructions from other repositories
| Mod | Type | Stars | Last push | Licence | Scan | Tokens | Works with |
|---|---|---|---|---|---|---|---|
| Unreal-MCP AGENTS.mdIvanMurzak/Unreal-MCP | Instructions file | 57 | yesterday | Apache-2.0 | A | 2,888 | Codex, OpenCode |
| vscode buildNext.instructions.mdmicrosoft/vscode | Instructions file | 195k | yesterday | MIT | A | 9,719 | GitHub Copilot |
| next.js AGENTS.mdvercel/next.js | Instructions file | 143k | yesterday | MIT | A | 7,461 | Codex, OpenCode |
| langchain AGENTS.mdlangchain-ai/langchain | Instructions file | 149k | yesterday | MIT | A | 4,680 | Codex, OpenCode |
| vscode oss-third-party-notices.instructions.mdmicrosoft/vscode | Instructions file | 195k | yesterday | MIT | A | 5,516 | GitHub Copilot |
| vscode copilot-instructions.mdmicrosoft/vscode | Instructions file | 195k | yesterday | MIT | A | 3,093 | GitHub Copilot |
| deepseek-harness AGENTS.mddeepseek-ai/deepseek-harness | Instructions file | 274k | 2d ago | MIT | A | 4,313 | Codex, OpenCode |