The Agent Skills Directory

[COMMAND_EXECUTION]: The Python module scripts/rbx_architecture.py facilitates the execution of external development tools through subprocess.run using list-based arguments. These commands are used to build Roblox place files via rojo, execute Luau scripts for offline auditing via lune, and perform static analysis via luau-lsp. These operations are standard for the intended Roblox development workflow and operate on local project files.
[DYNAMIC_EXECUTION]: The skill utilizes dynamic script loading and execution patterns within the Roblox ecosystem. Specifically, scripts/agentkit/architecture/Loader.luau manages the initialization of game services using the require function. Additionally, the lune runtime is used to execute Luau-based audit scripts externally to Roblox Studio.
[INDIRECT_PROMPT_INJECTION]: The skill provides an auditing surface in scripts/agentkit/architecture/ArchAudit.luau that parses user-authored Luau code. While this ingests untrusted data, the tool's behavior is restricted to pattern matching for security risks (e.g., hardcoded secrets or network requests) and reporting findings. It does not interpolate untrusted code into executable contexts for the agent or the system.
[SAFE]: The skill includes security-positive features, such as the ArchAudit tool which specifically scans for sensitive patterns like HttpService calls, hardcoded tokens, and secret assignments to help developers secure their projects. All external repository references target the official 'scenario-labs' organization, which is the verified author of the skill.

scenario-roblox-architecture