roblox-asset-library
Pass
Audited by Gen Agent Trust Hub on Oct 2, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONREMOTE_CODE_EXECUTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted third-party assets and metadata from the Roblox Creator Store and user inventories, which presents an attack surface for indirect prompt injection.
- Ingestion points: External asset metadata and hierarchy discovery through
search_assetand inventory resolution (SKILL.md, Step 3). - Boundary markers: Instructions mandate the use of an isolated
ServerStorage.__AssetReviewquarantine folder to prevent assets from interacting with production systems during the review phase. - Capability inventory: The skill possesses the ability to search, insert, and inspect instances within the Roblox DataModel, including the ability to read source code from
LuaSourceContainerobjects. - Sanitization: The skill enforces a rigorous static security review process, requiring the agent to identify and strip unauthorized
Script,LocalScript,ModuleScript, and remotes before integration (import-checklist.md). - [REMOTE_CODE_EXECUTION]: The skill manages the importation of Roblox models which may contain executable Lua scripts, representing a form of remote code delivery.
- Evidence: The
import-checklist.mdandSKILL.md(Step 4) contain non-negotiable rules against executing bundled code and provide technical procedures for identifying and removing hidden executable content. - [COMMAND_EXECUTION]: The skill utilizes specialized environment tools for scene analysis and performance profiling.
- Evidence: References to
$roblox-docs,$roblox-device-testing,$roblox-scene-analysis, and$roblox-performance-profilingindicate interaction with the Roblox development environment to verify asset integrity and runtime impact.
Audit Metadata