roblox-avatar-appearance
Pass
Audited by Gen Agent Trust Hub on Sep 20, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill retrieves and processes asset data from the public Roblox Marketplace API, which could potentially contain malicious instructions embedded in asset names or descriptions designed to influence agent behavior.
- Ingestion points: Fetches data from the catalog search endpoint at
https://catalog.roblox.com/v1/search/itemsas described inSKILL.md. - Boundary markers: The skill does not implement specific delimiters or instructions to treat the retrieved marketplace metadata as untrusted content.
- Capability inventory: Utilizes
WebFetchfor network access,InsertService:LoadAssetfor instantiating models, andscreen_capturefor visual validation of results. - Sanitization: There is no mention of sanitizing or validating the text-based metadata (such as names or descriptions) returned by the API before it is handled by the agent.
Audit Metadata