The Agent Skills Directory

[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface through its data processing workflow.
Ingestion points: The skill ingests untrusted data from the external Roblox environment via the get_console_output tool as described in the workflow section of SKILL.md.
Boundary markers: The instructions lack delimiters or explicit directives to ignore or treat console output as untrusted data, which could allow the agent to be influenced by instructions embedded in the console logs.
Capability inventory: The skill possesses capabilities to interact with the environment through user_mouse_input, user_keyboard_input, and character_navigation tools (SKILL.md), which could be misused if the agent obeys instructions from the ingested logs.
Sanitization: There is no evidence of sanitization, filtering, or validation of the content received from the console output before it is processed by the agent.

roblox-mobile-playtest