The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The documentation suggests installing the unity-mcp-cli package via npm (npm install -g unity-mcp-cli). This is a standard installation procedure for the command-line interface required to bridge the AI agent with the Unity Editor environment.
[COMMAND_EXECUTION]: The skill's primary function is to execute shell commands using the unity-mcp-cli tool. These commands are used to invoke the scene-set-active functionality within Unity, which is the intended purpose of the skill.
[INDIRECT_PROMPT_INJECTION]: The skill processes structured input via the sceneRef parameter to identify Unity assets. While this represents a surface for indirect injection (processing external data to perform actions), the risk is low as it targets specific Unity API calls (EditorSceneManager.SetActiveScene) and requires the scene to already be opened in the editor environment.
Ingestion points: The sceneRef object (containing instanceID, assetPath, and assetGuid) in SKILL.md.
Boundary markers: None present in the input schema.
Capability inventory: Execution of Unity Editor commands via unity-mcp-cli.
Sanitization: Not explicitly defined in the skill markdown; relies on the underlying CLI tool's validation of asset references.

scene-set-active