The Agent Skills Directory

[SAFE]: The skill facilitates standard Unity editor tasks for XR development. It incorporates extensive documentation, anti-hallucination tables, and explicit instructions to validate property names against verified XRI components.
[INDIRECT_PROMPT_INJECTION]: The skill presents an indirect injection surface through the ingestion of Unity scene data (object lists and reports) and the capability to wire event callbacks to arbitrary methods via reflection (xr_add_interaction_event). Boundary markers include instructional warnings to load schemas before use and verified property tables in API_REFERENCE.md. While no specific sanitization logic is described for method names, no malicious exploitation of this surface was detected in the skill instructions.

unity-xr