The Agent Skills Directory

[COMMAND_EXECUTION]: The skill relies on the mcp__ue58-mcp__execute_python_code and mcp__ue58-mcp__call_tool tools to execute logic within the Unreal Engine environment. These tools provide the agent with the ability to run arbitrary Python code to manipulate engine assets.
[DYNAMIC_EXECUTION]: The skill instructs the agent to dynamically generate Python scripts from templates and execute them at runtime. This includes operations like asset creation, component addition, and property modification using the unreal Python API. While these follow specific templates to ensure engine stability, the code is assembled based on the user's specific request.
[INDIRECT_PROMPT_INJECTION]: The skill acts as a bridge between user instructions and code execution within a high-privilege environment (Unreal Engine). It is susceptible to indirect prompt injection if malicious data is provided as asset names, paths, or property values.
Ingestion points: User-supplied strings for asset_name, package_path, component_type, and property_value across various workflows defined in SKILL.md and reference/common_workflows.md.
Boundary markers: The provided templates do not utilize boundary markers or explicit instructions to sanitize or validate user-provided strings before they are interpolated into executable Python code.
Capability inventory: The skill has significant capabilities, including the ability to load assets (unreal.load_asset), spawn actors in a level (EditorLevelLibrary.spawn_actor_from_class), and modify asset properties via the unreal module.
Sanitization: There is no evidence of input sanitization or validation logic in the provided reference scripts.

unreal-blueprint-automation