roblox-npc-ai

Pass

Audited by Gen Agent Trust Hub on Oct 15, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests game state data, such as player character positions and health, to inform NPC decision-making and pathfinding.
  • Ingestion points: Character data and player lists are retrieved via Players:GetPlayers() and FindFirstChild calls within the findNearestPlayer and canDetectPlayer functions in references/full.md.
  • Boundary markers: The skill does not explicitly use boundary markers for this data, as it is processed as numerical and object-based game state rather than natural language instructions.
  • Capability inventory: Capabilities are limited to Roblox engine movement functions, such as humanoid:MoveTo() and path:ComputeAsync(), which do not interact with the host file system or network outside the game environment.
  • Sanitization: Input data is used directly in mathematical checks (Magnitude, Dot product) and standard engine APIs.
  • [SAFE]: The skill promotes security best practices for Roblox game development by recommending the use of Server Authority and the SetNetworkOwner(nil) method to prevent clients from gaining unauthorized physics control over NPCs.
  • [SAFE]: External documentation links point to official Roblox creator domains and established developer community forums.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 15, 2026, 01:29 PM