Usage instructions
OpenClaw Unreal Skill is an AI-assisted development tool designed specifically for Unreal Engine 5. It establishes an HTTP communication bridge with the editor through OpenClaw Gateway, allowing AI to directly control the editor to perform various development tasks. This skill is built using TypeScript and runs as a local plug-in, providing game developers with a complete set of automated editor operation solutions.
Core usageaspect, this skill passes/unreal/*The endpoint communicates bi-directionally with the local Unreal Editor, supporting over 20 editor manipulation tools. Developers can use natural language instructions to let AI perform level management (open, save, switch levels), Actor operations (create, delete, modify objects such as Cube, Light, Camera, etc.), Transform adjustments (position, rotation, scaling), component management (add/remove components), and editor control (Play In Editor start/stop/pause). In addition, it also supports debugging functions (screenshots, hierarchical viewing, log output), input simulation (keyboard and mouse events), asset management (browse, import) and blueprint operations (list, open). All operations are implemented through the local HTTP polling mechanism, without cloud dependencies.
Significant advantagesIncluding: highly integrated workflow, AI can directly understand and execute complex editor operation sequences, greatly reducing manual repetitive work; localized processing architecture, all data is only transmitted between the local Gateway and Editor, and is not uploaded to third-party servers, ensuring project privacy; flexible scalability, open source based on the MIT protocol, developers can customize the tool set according to needs; complete error handling mechanism, including request size limit (1MB), JSON parsing verification, 60 Second timeout protection and automatic session cleaning (deletion without heartbeat within 2 minutes) ensure system stability.
Potential Disadvantages and LimitationsWorth noting: Currently in the Beta stage (v0.9.5), there may be undiscovered stability issues and it is not suitable for key production projects; it relies on a specific version of Unreal Engine 5.x, and its backward compatibility is questionable; the HTTP polling architecture is simple but slightly less real-time than WebSocket; although the AI automatic invocation function (disableModelInvocation: false) improves efficiency, it may produce unexpected operations in complex scenarios; it only supports local stand-alone development mode and cannot support team collaboration or remote editing scenarios.
Suitable target groupMainly include: independent game developers who need to quickly verify prototypes and automate repetitive level editing work; small game development teams who want to improve the efficiency of editor operations and reduce human errors; technical artists (TA) who need to batch adjust scene object properties or automate the asset import process; and Unreal Engine education practitioners who can be used to demonstrate editor automation and Python/Blueprint alternatives.
Risks of useIn terms of security, although the security rating reaches A level, users still need to pay attention: the Beta version may cause project data anomalies, and be sure to back it up before use; although there are no dangerous functions, AI can be executed directlyactor.deleteFor destructive operations such as this, instructions need to be verified carefully; although the local HTTP communication on port 18789 is limited to the local machine, it is still necessary to ensure that no malicious program is listening to the port; the version matching issue between the plug-in and the Gateway may cause the connection to fail; large-scale scenario operations (such as obtaining all Actors) may cause the request to fail due to excessive data volume (close to the 1MB limit).
Safety review
Core usage
OpenClaw Unreal Skill is part of the OpenClaw ecosystem and is used to remotely control the Unreal Editor via an AI assistant. After the user installs the accompanying Unreal Plugin, Skill establishes two-way communication with the Unreal Editor through the HTTP long polling mechanism to realize natural language-driven level editing, Actor operation, game debugging and other functions.
Main functional modules
|module|Typical operations|
|------|---------|
|**Level Management**|Open/save a level, get current level information, list all levels|
|**Actor Operation**|Create/delete Actors (Cube, Light, Camera, etc.), find Actors, modify properties|
|**Transformation Control**|Get/set position, rotation, scale|
|**Component Management**|Add/remove components, get component list|
|**Editor Controls**|Play In Editor (PIE), pause/resume, get editor status|
|**Input Simulation**|Keyboard, mouse, and joystick axis input simulation for automated testing|
|**Resource Browse**|Content Browser browsing, external resource import|
|**Debugging Tools**|World outline view, screenshots, log output|
|**Console commands**|Execute Unreal console commands and obtain output logs|
|**Blueprint Operation**|List blueprints and open blueprints in the editor|
Typical workflow
User: Create a cube at (100, 200, 50) AI: → Call actor.create → Unreal Editor generates Cube User: Take a screenshot to see the effect AI: → Call debug.screenshot → Return to viewport screenshot User: Start game testing AI: → Call editor.play → Start PIE mode
Significant advantages
1. Zero runtime dependencies: Use only Node.js built-inshttpmodules andopenclaw/plugin-sdkType definition, no third-party libraries introduce supply chain risks
2. local communication architecture: All HTTP endpoints (/unreal/*) Only listens to the local 127.0.0.1:18789, no outgoing network requests, and the data does not leave the local environment.
3. High functional integrity: 30+ tool interfaces covering the core development scenarios of Unreal Editor, supporting the entire process from level editing to PIE debugging
4. T2 Trusted Source: Affiliated to the OpenClaw community ecosystem, with supporting open source Unreal Plugin, complete change log, and semantic version management
5. AI native design: Automatic model calling is enabled by default (disableModelInvocation: false), AI can autonomously perform auxiliary tasks (checking levels, screenshot verification, etc.) to improve development efficiency
6. Flexible configuration:Support project levelopenclaw.jsonand overall~/.openclaw/unreal-plugin.jsonDual level configuration
Potential Disadvantages and Limitations
1. Version binding: Currently only Unreal Engine 5.x is supported, and UE4 compatibility is not declared; C++ plug-ins need to be installed manually, and the configuration threshold is increased for Blueprint-only projects
2. Editor dependencies: A large number of tools (Actor modification, level editing) require the editor to be in a non-PIE state, and the AI needs to understand the editor state machine to operate safely
3. name sensitivity: Actor search is case-sensitive matching, the user input "playerstart" may not match "PlayerStart"
4. Network mode restrictions: HTTP long polling mechanism may have delays in large projects or complex levels, non-WebSocket real-time connections
5. Blurred permission boundaries：disableModelInvocationDefault isfalseAlthough efficiency is improved, users need to clearly understand the meaning of AI being able to automatically perform operations such as deleting Actors and modifying attributes.
Suitable for the crowd
- independent game developer: Requires rapid prototyping verification and natural language driven level construction
- Technical Art (TA): Hope to batch adjust lighting, camera layout, and scene composition through AI assistance
- QA/Automation Test Engineer: Automated game testing with input simulation and PIE control
- Unreal teaching scene: Reduce the learning cost of editor operation and explore engine functions through natural language
General risks
|Risk type|illustrate|Mitigation measures|
|---------|------|---------|
|**Unexpected Asset Modification**|AI automatic calling may accidentally delete/change Actors|Save regularly and use version control|
|**Editor status confusion**|Trying to modify the level Actor in PIE will fail|Skill has done internal status checks, users should also pay attention|
|**Local service exposed**|CORS is configured as `*`. Theoretically, there is a cross-domain risk.|Only 127.0.0.1 is actually monitored, so the attack surface is extremely small.|
|**Input simulated interference**|Simulating keyboard/mouse may affect user operations|Recommended to use in independent testing window|
|**Session Expired**|The gateway needs to be reconnected after restarting.|Plug-in supports automatic reconnection|
Safety rating basis
The S-level rating is based on: zero third-party dependencies to eliminate supply chain attacks, pure local HTTP communication with no risk of leakage, T2 trusted community sources, and static/dynamic analysis with no high-risk findings. The two Low risks (CORS lenient, input validation simplification) are limited exposure issues on local service endpoints and do not affect the core security rating.
development-engineeringautomationproductivity
Copyright and takedown notice: AI Islands curates this page from public information. Skills, code, documents and packages remain the property of their original authors or rights holders. This listing is provided for indexing, research and installation convenience. If you believe any listing or download link infringes your rights, contact ai-islands@streamflowintel.com with proof of ownership, relevant URLs and your request. We will review and remove or adjust the content promptly. Review package permissions, dependencies and safety risks before installing.