The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes Unreal Engine's native command-line utilities, such as UnrealEditor-Cmd.exe and RunUAT.bat, to perform engine tasks like map validation and project packaging. These are standard local operations within an Unreal Engine development environment.
[DYNAMIC_EXECUTION]: The skill facilitates the generation and execution of Python scripts and Batch/PowerShell CI jobs. This is primarily achieved through the -ExecutePythonScript argument in the Unreal Editor, which is the intended mechanism for editor automation and pipeline integration.
[INDIRECT_PROMPT_INJECTION]: The skill involves parsing engine-generated logs (Saved/Logs/Log.txt) and screenshots to identify errors or visual regressions. This introduces an attack surface where untrusted content within log files could potentially influence agent reasoning.
Ingestion points: Unreal Engine log files (Saved/Logs/Log.txt) and visual output logs.
Boundary markers: None identified; content is parsed directly for error strings.
Capability inventory: The skill allows for the execution of batch scripts and editor-based Python scripts.
Sanitization: The skill focuses on specific error patterns (e.g., Error:, LogShaderCompilers) which limits the likelihood of accidental instruction following from log data.

unreal-engine