The Agent Skills Directory

[SAFE]: The skill implements expected functionality for Unity scene manipulation. It provides clear guidance on tool usage, identifies batch processing as a best practice, and defines clear targeting priorities (entityId, instanceId, path, name).
[SAFE]: The documentation for high-risk operations (e.g., gameobject_delete) explicitly mentions safety protocols and execution modes (NeverInSemi, FullAuto), which are standard security controls for agent platforms.
[INDIRECT_PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection as it retrieves data (names, tags, layers) from the Unity environment which could contain malicious instructions.
Ingestion points: Tools like gameobject_find and gameobject_get_info in SKILL.md ingest strings directly from the active Unity scene.
Boundary markers: The instructions do not define specific delimiters for separating environment data from agent instructions.
Capability inventory: The skill has significant capabilities to modify the environment, including creating, renaming, and deleting objects.
Sanitization: There is no mention of sanitization or filtering for data retrieved from the Unity scene.

unity-gameobject