roblox-luau-architecture
Pass
Audited by Gen Agent Trust Hub on Aug 2, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface because it ingests untrusted data from Roblox scripts and possesses tools to modify the environment. \n
- Ingestion points: The workflow utilizes
script_read,script_search,script_grep, andget_console_output(SKILL.md) to incorporate external script content into the agent's context. \n - Boundary markers: The skill does not provide specific instructions or delimiters to distinguish between architectural guidance and potentially malicious instructions embedded in read scripts. \n
- Capability inventory: The skill has access to
multi_editfor writing code andstart_stop_playfor executing it within the Roblox Studio environment (SKILL.md). \n - Sanitization: No sanitization or verification of the ingested code is implemented before the agent performs modifications. \n- [SAFE]: The skill advocates for critical security practices in Roblox development, such as ensuring game logic remains server-authoritative and validating all inputs received from client-side remotes.
Audit Metadata