The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: As a code generation and pattern guidance skill, this skill creates a surface for indirect prompt injection by defining how the agent should process and generate Luau code. \n
Ingestion points: The skill processes developer-provided code requirements and architectural constraints (SKILL.md, references/full.md).\n
Boundary markers: The skill does not explicitly instruct the agent to use boundary markers for external data processed by the resulting Luau scripts, though it does recommend routing networking and persistence to specific specialized skills.\n
Capability inventory: The skill itself contains no subprocess calls, file writing, or network operations across any of its files.\n
Sanitization: Not explicitly defined in this patterns skill, as it focuses on code structure and ownership rather than input validation.\n- [SAFE]: The skill promotes secure and efficient coding practices, such as explicit ownership for resource cleanup, proper error handling with pcall to avoid semantic collapse, and avoiding legacy functions that cause performance bottlenecks. All referenced sources are official documentation or well-known developer forums. The references to other skills (e.g., roblox-networking, roblox-data) indicate a modular architecture for handling sensitive operations.

roblox-luau-patterns