The Agent Skills Directory

[SAFE]: The skill provides procedures for auditing local Unreal Engine installations using read-only shell commands like grep and find. These commands are standard for development diagnostics and do not target sensitive user data or system configurations.
[INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to ingest data from local engine files, such as .uplugin descriptors and source code headers. While reading untrusted data is a theoretical vulnerability surface, the risk is considered safe in this context as the skill targets the local, legitimate engine installation required for the project.
[COMMAND_EXECUTION]: The skill includes benign shell command templates for counting occurrences of symbols or searching for plugin descriptors. These commands do not involve privilege escalation, network exfiltration, or persistence, and are intended for manual or agent-assisted project analysis.

ue5-feature-maturity-gate