The Agent Skills Directory

[SAFE]: The skill consists entirely of instructional markdown and configuration files without embedded scripts or executable code.
[INDIRECT_PROMPT_INJECTION]: The skill defines an attack surface by ingesting runtime data from Roblox Studio (e.g., console logs, scene data) which could theoretically contain instructions. However, this risk is managed by requirements for human review of visual evidence and explicit user authorization for data-writing or external services.
Ingestion points: The agent reads get_console_output, SceneAnalysis results, and the current Studio environment state (SKILL.md).
Boundary markers: No specific delimiters or boundary markers are defined for the ingested data.
Capability inventory: Available tools include start_stop_play, rbx-unit-test, rbx-perf-profiling, rbx-debug, and screen_capture (SKILL.md).
Sanitization: The skill does not explicitly mandate sanitization but enforces human-in-the-loop verification for visual results and technical pass/fail states.
[COMMAND_EXECUTION]: The instructions guide the agent in using platform-specific Roblox tools (start_stop_play, rbx-debug) within a controlled development environment. The skill includes explicit technical constraints, such as prohibiting DataModel modifications during debug pauses, to prevent unintended side effects.

roblox-game-qa-performance