UnrealSAFE
Enable AI assistant clients like Cursor, Windsurf and Claude Desktop to control Unreal Engine through natural language using the Model Context Protocol (MCP).
Overview
From the repository's own README, as read at the audited commit. Badges and raw HTML are left out.
unreal-mcp
[](https://opensource.org/licenses/MIT) [](https://www.unrealengine.com) [](https://www.python.org) [](https://github.com/chongdashu/unreal-mcp)
This project enables AI assistant clients like Cursor, Windsurf and Claude Desktop to control Unreal Engine through natural language using the Model Context Protocol (MCP).
⚠️ Experimental Status
This project is currently in an EXPERIMENTAL state. The API, functionality, and implementation details are subject to significant changes. While we encourage testing and feedback, please be aware that:
- Breaking changes may occur without notice
- Features may be incomplete or unstable
- Documentation may be outdated or missing
- Production use is not recommended at this time
🌟 Overview
The Unreal MCP integration provides comprehensive tools for controlling Unreal Engine through natural language:
57c99a9dc4fbOBSERVED · 2026-09-23Connect
Built from this server's own package name, version and transport as found in its source — not copied from anyone's documentation, so it cannot drift against a page we do not control.
claude mcp add unreal-mcp -- uvx unreal-mcp
{
"mcpServers": {
"unreal-mcp": {
"command": "uvx",
"args": [
"unreal-mcp"
]
}
}
}Exposed tools (30)
8 read · 21 write · 1 destructive. Blast radius: 1 tool can delete or overwrite — an agent that can be talked into calling a tool can be talked into calling this one.
| Tool | Risk | Description |
|---|---|---|
add_blueprint_event_node | write | |
add_blueprint_function_node | write | |
add_blueprint_get_self_component_reference | write | |
add_blueprint_input_action_node | write | |
add_blueprint_self_reference | write | |
add_blueprint_variable | write | |
add_button_to_widget | write | |
add_component_to_blueprint | write | |
add_text_block_to_widget | write | |
add_widget_to_viewport | write | |
bind_widget_event | read | |
compile_blueprint | read | Compile a Blueprint. |
connect_blueprint_nodes | read | |
create_blueprint | write | Create a new Blueprint class. |
create_input_mapping | write | |
create_umg_widget_blueprint | write | |
delete_actor | destructive | Delete an actor by name. |
find_actors_by_name | read | Find actors by name pattern. |
find_blueprint_nodes | read | |
get_actor_properties | read | Get all properties of an actor. |
get_actors_in_level | read | Get a list of all actors in the current level. |
set_actor_property | write | |
set_actor_transform | write | Set the transform of an actor. |
set_blueprint_property | write | |
set_component_property | write | Set a property on a component in a Blueprint. |
set_physics_properties | write | Set physics properties on a component. |
set_static_mesh_properties | write | |
set_text_block_binding | write | |
spawn_actor | write | Create a new actor in the current level. |
spawn_blueprint_actor | read | Spawn an actor from a Blueprint. |
Trust audit
SAFEgrade B · trust 89/100 Nothing in the source contradicts what it says it does. Grade A is reserved for packages that have also passed the behavioural sandbox.
| Layer | What it checks | Result |
|---|---|---|
| L0 | Provenance & inventory | PASS |
| L1 | Static analysis of the code | PASS |
| L2 | Instruction surface (what it tells the agent) | PASS |
| L3 | Class-specific surface | WARN |
| L4 | Behavioural (sandbox) | SKIPPED |
What the source does
- Filesystem
- declared (1 observation(s))
- Network
- declared (1 observation(s))
- Shell
- none-observed
- Dependencies
- pinned
- Secrets in source
- none-found
Findings (6)
delete_actor
.clinerules
.windsurfrules
.vsconfig
curl -LsSf https://astral.sh/uv/install.sh | sh
Gates applied: no_behavioural_pass, no_license.
57c99a9dc4fbfull audit observations/trust-audit/mcp-server/chongdashu__unreal-1.json · Report an issue / request a re-scanAudit history
Every audit this server has had. A grade with a past is a grade somebody is still checking.
| Date | Source | Verdict | Grade | Score | Change |
|---|---|---|---|---|---|
| 2026-09-23 | 57c99a9dc4fb | SAFE | B | 89 | source changed, verdict held |
Questions
What is the Unreal MCP server?
Enable AI assistant clients like Cursor, Windsurf and Claude Desktop to control Unreal Engine through natural language using the Model Context Protocol (MCP).
What tools does Unreal expose?
30 in total: 8 read-only, 21 that write, and 1 that can delete or overwrite (delete_actor). Every one is listed on this page with its risk.
Is Unreal safe to connect to an agent?
The audit found nothing in the source that contradicts what it says it does, and graded it B (89/100). Grade A is held back for packages that have also passed a sandboxed behavioural run, which is why a clean server reads B. Separately from the audit: 1 of its tools can destroy data, so scope the token you give it to what you actually need.
What credentials does Unreal need?
No credential environment variables were found in its source, so it appears to need none.
How does Unreal run?
It speaks stdio, so it runs as a local process your client starts. It is published on PyPI as unreal-mcp.
How current is this page?
The grade is for one exact copy of the source (57c99a9dc4fb), read on 2026-09-23. The repository is watched and re-audited when it changes.