The Agent Skills Directory

[SAFE]: The skill implements a robust framework for 3D building mechanics using industry-standard patterns for spatial partitioning (Octrees, Spatial Hash Grids) and structural physics.
[INDIRECT_PROMPT_INJECTION]: The skill contains an attack surface for indirect prompt injection via the networking layer in scripts/building-network-manager.js.
Ingestion points: External data is ingested through BuildingNetworkClient.onMessage using WebSockets.
Boundary markers: Data is ingested as structured JSON; however, there are no specific delimiters to prevent the interpretation of embedded natural language instructions if the agent were to process chat logs or custom piece metadata from the network.
Capability inventory: The skill possesses the capability to modify the 3D scene state, manage game objects, and transmit data over the network. It does not have access to sensitive file systems, shell commands, or persistent system storage.
Sanitization: The skill uses standard JSON.parse() for basic data validation, ensuring the input conforms to expected types for piece positions and types.
[EXTERNAL_DOWNLOADS]: The skill depends on the three library, a well-known and trusted industry standard for 3D graphics on the web. No other external or untrusted dependencies were detected.

building-mechanics