uemcp CLAUDE.md

uemcp CLAUDE.md is an instructions file for coding agents from ATDev-Inc/uemcp. It costs 1,125 tokens per session, scanned A, original, MIT.

A project instruction guide for UEMCP, an open-source server that lets Claude control a live Unreal Engine 5 editor through its built-in Python remote-control system.

In plain words
What is it for?
Use it when adding or reviewing UEMCP features, especially commands that send vectors, rotations, coordinates, or other values into Unreal Engine.
Why use it?
It keeps generated editor commands safe and prevents shared Python-session state from being overwritten. It also sets rules for building and testing command snippets.

Instructions file

Install

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.

agentmods
npx agentmods add instructions/atdev-inc/uemcp/claude-md
Clone the repo
git clone --depth 1 https://github.com/ATDev-Inc/uemcp

Wrote 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.

agentmods badge for uemcp CLAUDE.md

README.md
[![agentmods](https://agentmods.dev/badge/instructions/atdev-inc/uemcp/claude-md.svg)](https://agentmods.dev/instructions/atdev-inc/uemcp/claude-md)
Your own site
<a href="https://agentmods.dev/instructions/atdev-inc/uemcp/claude-md"><img src="https://agentmods.dev/badge/instructions/atdev-inc/uemcp/claude-md.svg" alt="Measured on agentmods" height="20"></a>
Per session 1,125 This file is loaded in full into every session.
When invoked 1,125 The same file — it is already loaded in full.
Security scan A 1 finding. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

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.

ModelPer sessionOnce invoked
Fable 5 $0.01125 $0.01125
Opus 5 $0.00562 $0.00562
Sonnet 5 $0.00225 $0.00225
Haiku 4.5 $0.00112 $0.00112

Measured 3d ago against content hash 9a06b1e92aa5, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

uemcp CLAUDE.md scanned grade A with 1 finding 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 3d 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.

Makes network callslowCapability

Not a fault in itself. Listed so you know the mod talks to something, and to what.

- **Never add an Unreal-side plugin or non-stdlib runtime dependency.** "Nothing to install inside Unreal" is the whole premise; the only runtime dep is `mcp`. Asset providers in `assets.py` are stdlib-only (`urllib`/`zi
CLAUDE.md · 42 lines

How it starts

The opening of the file, as written. The whole thing — 42 lines — stays where its author put it; the contents beside it link to each section on GitHub.

UEMCP

Open source MCP server that lets Claude drive a live Unreal Engine 5 editor with no plugin to compile. It speaks UE's built-in Python remote execution protocol (UDP multicast discovery + TCP command channel), builds Python snippets, runs them in the editor's shared Python session, and parses structured JSON back out.

Key rules

  • Snippet interpolation MUST use repr() ({value!r}), never raw f-string interpolation. This is the security boundary that keeps tool arguments from escaping into the editor's Python session as executable code. Reviewers reject raw interpolation.
  • Prefix every in-snippet variable with _ (e.g. _target, _LABEL). Remote execution shares ONE Python session inside the editor; unprefixed names trample user state.
  • Snippet builders in snippets.py are pure functions (args -> Python source string): no I/O, no Unreal import. tests/test_snippets.py compile()s every builder, so any new builder needs a compile case there or CI cannot catch its syntax errors.
  • Reject non-finite floats for any numeric snippet input: route vectors/rotations/coords through _finite / _vector / _rotator. repr(inf)/repr(nan) emit bare inf/nan tokens that NameError inside the editor.
  • Never add an Unreal-side plugin or non-stdlib runtime dependency. "Nothing to install inside Unreal" is the whole premise; the only runtime dep is mcp. Asset providers in assets.py are stdlib-only (urllib/zipfile, with zip-bomb guards) and unit-tested with the HTTP layer monkeypatched.
  • A tool docstring is the model-facing spec. Claude reads it to decide whether/how to call the tool; state what params mean and which conventions apply (see below).
  • Raise RuntimeError with actionable messages ("No actor found with label X") instead of letting None flow into a confusing AttributeError. Wrap newer engine APIs in try/except with a fallback (support vanilla UE 5.0-5.7).
  • Adding a tool touches 5 places, keep them in sync: builder in src/uemcp/snippets.py, registration in src/uemcp/server.py, compile case in tests/test_snippets.py, reference entry in docs/tools.md, and a real-editor smoke test naming the engine version in the PR.

Read the full file on GitHub · 42 lines

Changes

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.

  1. 3d ago First seen · 42 lines · 1,125 tokens per session scan A 9a06b1e92aa5

Subscribe to this mod's changes

uemcp CLAUDE.md is an instructions file published in the GitHub repository ATDev-Inc/uemcp (0 stars, last pushed 11d ago), licensed MIT. It adds 1,125 tokens to every session, about $0.0056 per session on Opus 5. A static security scan graded it A with 1 finding (makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.