The Agent Skills Directory

[Prompt Injection] (SAFE): No adversarial instructions or jailbreak attempts were found in the skill definition or reference files.- [Data Exposure & Exfiltration] (SAFE): No hardcoded credentials, sensitive path access, or unauthorized network activity detected.- [Unverifiable Dependencies & Remote Code Execution] (SAFE): No download or execution of external scripts. The skill specifically includes a validation rule ('check-require-module-id') to prevent the use of untrustworthy remote Roblox modules.- [Indirect Prompt Injection] (SAFE): The skill includes an ingestion surface for user-provided code, but provides strong mitigation through its 'validations.md' reference file, which checks for common security pitfalls like missing server-side validation and unsafe HTTP usage.- [Dynamic Execution] (SAFE): No use of dynamic code execution or runtime compilation. The skill provides linting-like rules to ensure the safety of code generated for the user.

roblox-development