ue-test-authoring
Pass
Audited by Gen Agent Trust Hub on Sep 4, 2026
Risk Level: SAFECOMMAND_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses the
Bashtool to perform environmental checks, specifically searching for a.uprojectfile in the working directory to ensure it is operating within a valid Unreal Engine project. - [DYNAMIC_EXECUTION]: The skill enables the agent to compile and link the Unreal Engine project (
build_solution_start) and execute Python commands directly within the live Unreal Engine editor (ue_execute_python). These actions involve generating and running executable logic, which are core requirements for its stated purpose of automated test authoring. - [INDIRECT_PROMPT_INJECTION]: The skill reads and processes project files and source code, creating a potential vector where instructions embedded in untrusted codebase files could influence the agent's behavior.
- Ingestion points: Reads the
.uprojectfile, module configuration files (Build.cs), and various C++ source files (.h,.cpp) to understand the project structure and existing test patterns. - Boundary markers: No specific delimiters or warnings are used to distinguish project code from agent instructions.
- Capability inventory: The skill possesses the ability to write and edit files, execute shell commands, build the project, and run scripts in the editor environment.
- Sanitization: There is no documented validation or filtering of content ingested from the project's source files.
Audit Metadata